Skip to main content

Personal Intelligent City Accessible Vehicle System

Final Report Summary - PICAV (Personal Intelligent City Accessible Vehicle System)

Executive summary:

For a better comprehension of the contents, please refer to the pdf document that exposes the complex concepts by their related images.

The project presents a new mobility concept for passengers ensuring accessibility for all in urban pedestrian environments. The concept addresses a new Personal Intelligent City Accessible Vehicle (PICAV) and a new transport system that integrates a fleet of PICAV units. The PICAV vehicle ensures accessibility for everybody and some of its features are specifically designed for people whose mobility is restricted for different reasons, particularly (but not only) elderly and disabled people. Ergonomics, comfort, stability, assisted driving, eco-sustainability, parking and mobility dexterity as well as vehicle/infrastructures intelligent networking were the main drivers of PICAV design. The innovative electric vehicle presents new unconventional chassis and suspension structure, new seating sub-assembly, new efficient power supply module.

Project Context and Objectives:

Project context
The PICAV vehicle because of its small dimensions, tiny footprint, on-board intelligence and zero environmental impact is suitable for moving in pedestrianized areas and it behaves almost as a pedestrian, disturbing other pedestrians as little as possible. Moreover, the PICAV system integrates itself with the existing public transport system to help it become more accessible for older and disabled people by acting as a smooth link between walking, bicycle and conventional public transport and thus extending the availability of a public motorized transport service to cover the entire urban area.

The PICAV transport system is an on-demand system and it is based on the car-sharing concept. To overcome the barriers of traditional car-sharing systems, the following specific services are provided: instant access; open-ended reservation; one-way trips and easy to find. The stacks of available PICAV vehicles will be conveniently placed in locations all around pedestrian areas close to interchange points with public transportation (like bus stops, subway, etc.): people can commute into the city, get off the train, jump in a PICAV vehicle, visit attractions (shops, monuments, museums) in the pedestrian area, drop the vehicle off at any PICAV parking lot on the order of the pedestrianized area and proceed their journey in the surrounding area again by public transport. An example of a possible application field for the proposed transport system refers to the historical city centre of Genoa using the connections with the traditional transport modes (bus, train, boat) as well as the information exchange between the PICAV vehicles and the existent ITS systems. The PICAV prototype was used within the area of the 2012 International Boat Show (see http://www.youtube.com/watch?v=Y9ElKorNjgM online) and within the historical city center of Genova (see http://www.youtube.com/watch?v=Y9ElKorNjgM&feature=plcp online).

The PICAV transport system joins the benefits of public transport (one-way trips, cost efficiency) to the ones of private car (security issues, flexibility) in areas where neither buses nor cars can operate (historical city centres, exhibition areas, parks, etc.).

The number of vehicles in a city's fleet is kept low, and can be determined by the public administration of the city, proportional to the served area and the extent that PICAV booking is priority based (emergency, disabled, elderly, tourist, etc.). The access to the service is thought by means of a smart card. The location of the nearest vehicle can be indicated to the user, together with directions from their current location to make it easier to find an available vehicle.

The project is based on the concept that in order to improve the quality of urban life in terms of congestion, pollution, accessibility etc., it is necessary to rethink the relationship between the car and the city's historical and restricted traffic areas and to propose urban personal transportation by providing the entire city's inhabitants better opportunities to practice 'good mobility' citizenship. Further areas of interest in using personal vehicles like PICAV are the many small villages located on the top of hills where the old (medieval) infrastructures are narrow gravel roads with steps. These kinds of urban ancient nucleus are typical of Europe.

Project objectives

The two main project objectives addressed by the call are:
- Radical move towards clean energy efficient, safe and intelligent personal vehicles. To this objective a number of new concept vehicle modules (sub-assemblies) that are considered critical for the targeted elderly and limited mobility end-users and for the environment, were developed and breakthrough solutions were proposed about: Ergonomic seat-module: for the comfort and well-being of users. Reactive suspension-module: to increase the safety of the vehicle. Clean efficient power-module: for greenhouse emissions, having as minimum requirements those targeted in the Kyoto and Euro V agreements. Driving assistance-module: for collision avoidance, parking within crowded roads, etc. Body embedded sensorial system: aimed at providing a safe behaviour within crowded pedestrian environments. Intelligent vehicle-infrastructure networking: so that the vehicle can respond adequately to real-time changes in the network status integrating itself to the public transport network already existent with the urban area.

- Paradigm change in user service aiming at an efficient organization of the mobility of weak or for some reason hampered people in urban areas. The new vehicle, beside the private use, is thought as a unit of a small car sharing fleet that can be rented on demand. The rational use of the fleet is guaranteed by specific management strategies and is based on the knowledge of the status of the entire fleet and on the existing networked public infrastructures. The rational use of the PICAV fleet is pursued through: Optimization of the fleet dimension: for a given transport demand and for a given served area/infrastructure. Choice of suitable PICAV parking lots: encouraging inter-modal transport chains and accessibility. Automatic elaboration of routes: helping the users to plan their travels. Definition of management strategies: aiming at keeping the distribution balance of parked PICAV vehicles between parking lots. Development of a dynamic simulator: able to reproduce and to test a given PICAV transport system and its management strategies. The efficiency of the transport system is related to its capacity to satisfy the transport demand with the desired level of service (issues like waiting times at the parking lots and interchange points will be taken into account) keeping acceptable the negative impacts on the non-users.

In order to achieve these high level objectives the following specific objectives were defined.
- Size and Weight: all the dimensions of PICAV vehicle prototype were kept as small as possible but compatible with ergonomic user needs and environmental infrastructure. The partnership tried to keep the vehicle mass as low as possible.
- Agility: Dexterous operation in narrow roads and short corners; ability to overcome steps up to 200 mm; ascend and descend a maximum longitudinal incline of 25 degrees; a maximum tilt of 25 degrees without toppling; turning radius smaller than 1m.
- Eco-Compatibility: the new transport system provides transportation facilities in areas of the cities where no public transport is available and in most cases no private transport means can be used at all. This has to be achieved without introducing any pollution, thanks to the propulsion system used by the vehicles, and with a minimum noise impact. Targeted measurable objectives are: - zero emission; - noise emission less than 45 dBA (for comparison, the legal maximum noise level for a small motorcycle is 77 dBA and 45 dBA is about the level of a quiet urban area without traffic of any kind and with few pedestrians walking around).
- Personal comfort and Ergonomics: the vehicle components and sub-assemblies were tested by representative sample of potential users. They were be asked to sit in the vehicle, to drive it in a given scenario and to give feedback. The level of satisfaction about comfort and easy driving also in transient conditions (breaking, turning, acceleration) was good.
- Safety: the vehicle maximum velocity is lower than 6 m/s thus no crash tests are foreseen. Stability is an important issue for safety. With reference to the agility performances, the vehicle guarantees roll and tilt in an angular range of ±12 degrees. It was verified that the user's exit from the vehicle is possible in case of overturning in any direction. Safety issues were addressed considering robustness of the control and redundancy of sensing, actuation and control systems. The design of the vehicle exterior ensured that the surfaces are smooth with no protuberances that could injure passing pedestrians.
- Energy efficiency: The power system architecture and specific algorithms were developed to improve the vehicle energy efficiency. It resulted in more than 25% with reference to current urban transport means.
- Transport service: in principle, the benefits of the proposed transport system on the transport service are not quantifiable because nothing similar exists and because of the social value of the service to a category of people usually excluded by the public transport services. The performance of the transport service logics was assessed by the following indicators: accessibility as a measure of availability, distance between PICAV parking lots and public transport stops at the border area less than the maximum walking distance for elderly; transport demand as a measure of quality: in general the identified potential users judged satisfactory the simulated transport system (virtual reality testing); waiting times as a measure of reliability: waiting time at the PICAV parking lots comparable with the waiting time at the public transport stops at the border area.

Project Results:

The main scientific and technical results are summarized in the two following sections concerning respectively the PICAV vehicle prototype and the PICAV transport system simulator.

1 PICAV vehicle prototype

1.1 Power system

The power system is the responsible for powering all the electric devices.
The battery system is made up of Lithium-polymer cells. Each cell has a nominal voltage of 3.2V and its capacity is 5Ah. The nominal current is 5A. The battery pack configuration is made by 15 blocks connected in serial and each block has 16 cells connected in parallel. When the battery is fully charged, the voltage is 54V. Nominal voltage is 48V and cut off voltage is 45V.

After having examined and tested a configuration with Li-ion cells for normal operation and a supercapacitor for supply peak power, it was decided not to use this configuration because the amount of cells was very high, the space for the supercapacitor was limited and it could only be charged when descending slopes and when it was fully charged could only deliver maximum peak power during 3 seconds. So, technology was changed to Li-Poly (Lithium Polymer). With this configuration we could storage all the energy in a volume of 20 litres with a weight of 40Kg.

Lithium batteries have a good performance but they need to be managed properly in order to make them last and to avoid dangerous situations. For this reason a BMS (Battery Management System) is included with the battery. This BMS is responsible for monitoring the state of the battery pack (current, voltage and temperature) and disconnect the system if critical conditions occur. It is responsible for managing cell balancing, a technique that extends battery life, transferring energy from the most charged cell to the less charged. BMS has also CAN communication features, in order to provide information about SOC and SOH of the battery to the system. A CAN protocol was also developed to standardize messages. As the BMS has a microcontroller inside, improvements can be made upgrading the software.

1.2 PICAV traction
The traction of the vehicle is made by actuators directly attached to the wheels, not entirely the same, but very similar to the new in-wheel motors available in the market. However, this solution represents a more economical alternative. The vehicle has 4 driving wheels. Each couple of side wheels are connected to the ends of a side link hinged on the chassis frame. This innovative configuration allows the vehicle to climb steps.

1.3 Steering system (2 motors)
The steering system is located in the rear side of the vehicle and has been realized with two independent motors: one on each wheel. To give the necessary torque for the wheels to turn when the vehicle is static, a gear system placed between the motor and the wheel steering frame is used, this gear system increases the torque to the minimum required to overcome the friction between wheel and terrain.

The steering system comprises different electric and mechanical devices that combined together permit the PICAV wheels to turn in a 180º angle, this, taking into account the necessary cabling and sensorial devices to control the turn. At the same time, the wheels being mirrored one respect to the other rotate with the same angle but in different direction.

The PICAV vehicle, thanks to the steering mechanisms employed, is capable of making a 360º turn almost in place, the turning radius is of 1m with the centre of rotation on the front wheels, and distance measured between the front and the rear wheels. During the design of the vehicle, and given the fact that the four wheels are powered, the steering mechanism could have been avoided and an even smaller turning radius would have been achieved. However, it was decided to adopt it in order to have a faster response during actual running of the vehicle and to decrease the complexity of the low level control system.

Further, the implemented solution of the steering mechanism represents as well a more efficient solution in terms of battery consumption.

1.4 Pitch control system (1 motor)
The pitch control system has been realized with one motor controlling the hydraulic system. The motor actuates a bi-directional hydraulic pump, which in turn modulates the flow of the fluid in the pressurized circuit for the pitch movement. It is done by actuating 2 hydraulic pistons (one for each wheel module), the pump sends the fluid to the upper chamber of the piston, pushing the piston shaft downwards realizing then a forward rotation of the body. When the desired movement is the backward rotation, the pump inverts the direction of flow, sending it to the bottom chamber of the piston and pushing it upwards.

As described above, the forward rotation allows to maintain horizontally the body of the vehicle when it goes downhill or coming down from the stairs. The backward rotation allows to maintain horizontally the body of the vehicle when it runs uphill or going up stairs.

1.5 Roll control mechanism (2 motors)
The roll control mechanism of the PICAV vehicle is designed in a way so that the vehicle is capable of adjusting the height of the wheel module with respect to the terrain. It is used as a novel suspension system actuated by electric motors. The vertical sliders comprise an electrical screw axis with motor arranged vertically, two rails with a pair of linear guides and two sliders on each are fixed in the C- structural beam to which the wheel unit is fixed at a pivoting pin. This permits the PICAV body to be lifted from the ground maintaining the two wheel modules parallel to each other (same height), this helps the user to level the preferred height in which they want to circulate and also if the terrain is known to be uneven can be set in advance a predefined height.

The advantage of having the slider mechanism embedded in the PICAV is far from giving only the possibility to move up and down the vehicle depending on the height the user wants to drive in, or according to the roughness of the terrain. Taking into account the urban scenario of Genova and other scenarios in which part of the city is atop a hill, it has to be considered that the PICAV will circulate in roads with a lateral incline. The roll mechanism permits the vehicle to control the height of each wheel module independently; this means that with a lateral incline the body of the vehicle maintains horizontality and become virtually impossible to turn over.

Furthermore, within the aim of this project, the vehicle needs to be capable of circulating in different environments including pedestrian areas. PICAV is capable of going with one wheel module above the sidewalk and the other below on the road; while, at the same time maintaining the driver in a horizontal position.

1.6 Seat mechanism (2 linear actuators)
The seat is strictly integrated in the vehicle's body, seat geometry and functions are strongly dependent on the vehicle's design. On the other hand, the seat is an independent result of the project and it has wide applicability beyond PICAV. For this reason, the functionalities required for integration in the vehicle have been separated from the components and functionalities which can have a stand-alone application.

The seat comprises a double frame, two armrests (integrating emergency stop, control buttons and steering joystick), two electric actuators and the seat module.

The two parts of the frame (base and pad) are hinged in the front, with axis of the hinge orthogonal to the seat symmetry plane and positioned to cross the user in the region of the knees (when the user is sat). In this way the rotating part of the frame (the pad) accommodates the rotation of the user knees when the user stands up and sits down. The base of the frame is a lightened aluminium shield with connections at the sides for the sliders used to realize seat translation in the vehicle.

The seat module has been designed taking into account the state of the art on ergonomics of automotive seats and the needs of impaired user. It envelops the user at the shoulders and legs; while the size ensures comfort for users of any height and size.

The motion of the seat is delivered by two linear actuators directly attached to it. A first linear actuator enables the forward movement of the seat, translation inside the vehicle. The second one starts to work when the seat is completely out of the vehicle; it raises the seat causing it to rotate on a hinge located on the front of it. When the user is leaning in the seat, the procedure is reversed accompanying the user in the inner drive position. This gives the possibility to disable user to be carried inside the vehicle in an autonomous way without having to use their muscles, which indeed might be unable to use. The actuation of this motion is commanded by the user by the buttons located in the armrest. The procedure is autonomous together with the cover opening and closing that will be described further on. This gives a huge advantage, by one touch of a button the user is carried inside the vehicle and the cover closed.

Moreover, the adaptation of the seat to the body of the user is realized at two levels:
a)using passive foam (traditional way);
b)with shape memory foam.

Passive foams are the common padding in seats and cushioned products like sofas and rest-chairs. They have been used to cover the metal skeleton of the seat and as additional finishing at the sides of legs and shoulders. In car seats these are the only material used for padding while in the PICAV seat they represent a very minor fraction of the whole seat padding.

1.7 Cover mechanism (2 linear actuators)
The cover of the vehicle is made in polycarbonate material. It has been realized embedding a motorized system that enables its opening with a rotation on a joint situated in the rear of the vehicle. In order to enable a comfortable entering and exiting of the user, two linear actuators, controlled by a logic unit, work together allowing opening and closing of the cover.

1.8 Braking solution (1 motor)
The PICAV integrates in its four wheels the brake callipers as those of a normal vehicle. However, they are used for emergency purposes only. A normal car pump actuated by another electric motor is used. The motor is connected to an emergency port of the battery, meaning that the last resources of the battery are used for it.

In normal operation, the vehicle brakes thanks to the traction motors, the user simply needs to pull the joystick back decreasing the power requirement of the motor and hence braking.

1.9 Control system
The control scheme proposed in this work considers sensor information from laser, odometers and inclinometers and the instruction from the HMI modules. The PICAV prototype is a complex system with many different actuators. For this reason has been defined a CAN open protocol to control them.

Sensors. Different sensors are used to provide driving assistance and a comfortable control. For the PICAV vehicle 3 front laser are used for the obstacle, steps and stair detection. Moreover, 4 ultrasonic sensors and a vision camera were used for the rear detection in the back driving maneuvers. Two inclinometers placed in the pitch and roll axle of the vehicle were used to keep the comfort in the driving. Finally, two inductive sensors were placed for the initialization of the slider and the hydraulic pump on the PICAV.

The stair detection system is composed of three laser range finders. The two laser sensors dedicated to the step detection have a vertical orientation in order to scan the altitude profile of the environment over two lines of sight. For the obstacle detection, another was placed in the middle of the vehicle in horizontal position. The system has been designed for the PICAV vehicle, which is supposed to be able to cross over a stair and steps.

The obstacle detection algorithm is divided in five stages. The first stage is in charge of the data acquisition from the lasers. At second stage a simple approach for the segmentation problem has been implemented using a recursive line fitting algorithm as it is classically employed for obstacle detection. Then the segmentation stage is to provide a set of well separated segments. The clustering stage uses the information from the segmentation process. The classification aim is to provide a type of object detected by the perception system. In this implementation different object, as well: for-wheeled vehicles (car), two-wheeled vehicles (2w), Pedestrian (ped) and Unknown (unk) can be detected. To this end, the work assumptions, as different sizes for classes, are considered. Finally, the tracking stage provides successive estimations of obstacle state, specially the obstacle speed, positions and orientation.

In stair detection algorithm original data are first projected on Z-axis. Only XZ-coordinated are considered to calculate derivative altitude. An algorithm is then implemented to detect peaks in the derivative altitude curve and finally, data coming from the two laser sensors are merged in order to provide, in a robust manner, an estimation of the stair: position, height and orientation.

Peak detection in a noisy environment is a well known problem in signal processing. In our step detection application, peaks correspond to changes in the orientation of the ground. A peak maximum corresponds to the top of the step and beginning of the peak to the base of the step. With this information, the system can assess the situation, the location, height and orientation of the step are sent to the vehicle which decides to pass over the step or not.

On board PC. All perception algorithms, control maneuvers and the driving assistance system run in a high level PC. The model used in the LS-377Q, a new generation of the 3.5 mini board, supports Intel Core i7. Different ports (serial, Ethernet, USB and CAN) were considered in the interconnection of the system. A first approach uses the CAN-Open protocol to communicate the actuators. The control has been done with devices for low level control, a then communicated with the on board high control PC through an Ethernet connection.

A first low control system uses the control unit named µRMC2 (AS1017.002) motion controller. This unit allow to control axis for motion control systems through devices that are capable of controlling a CAN Open network with different kind of brushless drives.

1.10 Driver assistant and HMI
The human machine interface is in charge to transmit the command from the user to the vehicle.
The HMI developed allows three main tasks: joystick, buttons and monitoring. The software is installed in an Tablet. The onboard PC has installed a Wifi-card, and through this connection the command can be read for the high level control program. Finally, other PC is used as monitoring station, it means, the HMI can run in different external PCs that have access in the access point. However, the simultaneous use of each task is prohibited.

For example, is the Ipad is using the joystick, the other PC is able to use just the button and the monitoring interface. Functionality of HMI:
Html web interface. The html web interfaces are widely used in the development of web pages, computer software and multimedia. For this application different design parameters were considered for the PICAV implementation: usability, visualization, functionality and accessibility. One of the advances of the html programming is that it can run at any internet browser. An RTMaps component was programmed to communicate the command with our high level control program.

Joystick. A touchscreen of the smartphones has been used to control the steering and the speed on the PICAV. However, from a standard computer (with a mouse) the joystick can also work properly. A low past filter was added to this reference signals or order to avoid jumps or peaks to the PICAV actuators. When the user lifts the finger from the touchscreen of the HMI, the filter smoothes the signals, and it sends the default value to zero.

Buttons. 6 buttons are used by the HMI to send setpoints to the low level of the PICAV. The different tasks defined are: Chassis (up and down); Cover (open and close); Seat (forward & backward); Seat (up and down); Ligths (on and off).

Monitor. variables monitored by the HMI from the information coming from the low level of the PICAV. The different variables read are: Actual speed Actual steering angle; Reference speed (slider); Reference steering angle (slider); Distance to the obstacles; Distance to the steps; Distance range to the back obstacles.

Driver assistant functions
These functions allow to aid drivers in critical situations more than to replace them. The main driver assistant functions developed by PICAV in urban scenarios are:
- Perception system;
- Remote functionality;
- Control driving.

The designed safe Human-Machine Interface has increased car safety and more generally road safety.

2 Simulator description
The present section of the report will display: the main characteristics of the proposed transport system, developed jointly by UNIPI and UCL, the main characteristics of the simulator, developed by UNIPI, and the main characteristics of the simulator GUI, developed by UCL. Herewith just a brief description of the transport system, of the simulator and of the optimization algorithm is reported. In deliverable D4.5. Simulator final release, a full description, method by method, of the simulator and of the optimization algorithm, is described.

2.1 The main characteristics of the proposed transport system
Three system management strategies have been developed. They overcome some restrictions of traditional car-sharing systems (first generation) where members need to book cars beforehand and the time the car will be dropped off should be specified (fixed-period reservation); besides, cars must be returned to the same parking lot where they were picked up (two-way trips).

The first two management strategies proposed in the PICAV transport system belong to the second generation car sharing systems. These systems give more flexibility to users providing:
- Instant access: users can access directly to an available vehicle, without the need to make a reservation;
- Open-ended reservation: users schedule the pick up time but can keep the PICAV vehicle as long as needed;
- One way trips: users can drop the vehicle off at any parking lot.

For these systems the risk to become unbalanced with respect with the number of vehicles at the multiple parking lots is high: due to uneven demand, some parking lots during the day may end up with an excess of vehicles whereas other parking lots may end up with none. According to the literature, there are two main categories of relocation strategies: user-based and operator-based, with some systems using a mixture of both. Operator-based relocation strategies resolve the balance problem through a strategy where some operators manually relocate a vehicle or a platoon of vehicles from parking lots having too many vehicles to parking lots having too few. Several transport systems of this typology have been realized in research pilot programs, for example the Coachella Valley in California (Barth and Todd, 1999), near Palm Spring's airport, the two CarLink experiences, IntelliShare, established at RiverSide University campus, the Praxitèle in Paris, the Honda ICVS in Singapore, then also CarLink I and CarLink II. The operator based relocation reports very high staff costs and some pilot applications, such as the Honda ICVS in Singapore, have turned out into a failure because of these too high costs. User-based relocation strategies ensure that at least part of the relocations is performed by users. The most important issue regarding the user-based strategies is the possibility of giving users some incentives, such as a reduction in prices, in order makes them improve the transport system performance.

According to Ciari et al. (2009), a higher degree of diffusion of vehicles within the area can better capture all the potential demand by making the system more competitive and more similar to private car, ensuring a high level of user satisfaction. This is the main idea of the third generation car-sharing systems where vehicles are available to users from any point of the intervention area: the third car sharing system proposed in this paper belongs to this generation. A high degree of capillarity results in a less amount of space needed for the parking lots, since some vehicles, as stated before, are not placed at the parking lots but along the roads; however, increasing capillarity increases the relocation costs. The Car2go is a transport system of the third generation and has been implemented in the city of Ulm (Firnkorn and Müller, 2011). In the Ulm car-sharing system, users are provided with a map of the intervention area, where they know in real time the exact position of all available vehicles. Users may book, require and check-in vehicles via a cellular GPRS technique; then they reach the vehicle by foot or any other mean of transport. When they have finished their trip, users can return the vehicle at any point in the intervention area. Surprisingly, in this reality of Ulm relocations do not result necessary and this is probably due to a well balanced transport demand and/or to a very high number of vehicles involved.

2.2 The PICAV vehicle
Details about the PICAV vehicle are displayed in the other sections of this report. Details about the battery and the electric engine, performed by Mazel, are also displayed.

The battery discharging law depends on: the average up and down slopes of the path, the PICAV speed, the air density, the PICAV frontal area, the aerodynamic and rolling coefficients, the mechanical and electrical efficiency, the motor voltage and the overall weight of the vehicle.

The battery charging technique is the opportunity charging. The term opportunity charging refers to the charging of the batteries wherever and whenever power is available. We define minimum charge level the quantity of charge necessary to the vehicle to perform the longest trip in a given environment. When a vehicle is returned by a user a check on its battery level is performed. If a vehicle has a level of charge below the minimum charge level, it needs to be recharged in the nearest parking lot. The battery charging time depends on initial charging level, final charging level and balancing time.

The PICAV speed is a function of pedestrian density along the street. Herewith are reported the two relationships between the PICAV speed and pedestrian density depending on driving mode.

The two equations are:
PICAV user driven: v = - 1.44677 k + 1.57751 1.1.
PICAV automatically driven: v = - 1.44677 k + 1.37751 1.2.

Where:
v = PICAV speed [m/s]
k = pedestrian density [ped/m2]

In the case of a person driven PICAV, its speed-density relationship is the result of a linear regression performed on empirical data collected in the historical city centre of Genoa (Cepolina et al. 2011). If the PICAV travels in a fully automatic way, it is envisaged that it travels at a speed which guarantees safety in dynamic pedestrian environments, given the perception and control systems PICAVs are provided with.

2.3 The proposed car sharing system
The first system management strategy does not require vehicles to move in a fully automatic mode. The other two management strategies require instead vehicles to move in a fully automatic mode and one of the most significant obstacles to the possible application of these systems is the fact that at the present automated cars are illegal on most public roads. However many countries are considering the legalization of automated cars and in a next future also the second and third system management strategies could be applied in the field. In the proposed car sharing system users are not assumed to know the positions of PICAV units. Thanks to smartphones this could be possible and some users, checking in real time the exact positions of the available PICAVs, could move towards them. However this option has not been taken into account in this research.

The first system management strategy
Vehicles can be accessed only at parking lots and PICAVs need to be driven by a person (they do not move in a fully automatic way). A fully user based relocation strategy is proposed. A system supervisor is in charge of addressing a subsection of users, who are called flexible users, to specific parking lots. A flexible user is a person that has a final destination outside the pedestrian area and reaches it by public transport; therefore a flexible user has a set of parking lots that are equally suitable for returning the PICAV unit. We call this a user's choice set. It includes all the parking lots close to inter-modal exchange points where public transport services, which are suitable to reach the user's final destination, stop. The flexible user agrees to return the PICAV unit to any of the parking lots within their choice set, because from these parking lots they can easily reach their final destination.

When a flexible user finishes their mission in the pedestrian area, they call the system supervisor and asks where they have to return the PICAV unit. The supervisor makes a decision according to the following information:
- The choice set of the flexible user;
- The current waiting times at the parking lots within the user's choice set;
- The travel times between the user's current position and the parking lots within their choice set.

The rules of the assignment are described in details in Cepolina and Farina (2012). The assignment results favorable from the point of view of the flexible user (since the assigned parking lot belongs to the user's choice set) and favorable from the point of view of the transport system management. The flexible user is supposed to return the PICAV unit to the parking lot chosen by the supervisor.

This management scheme is best suited when a consistent number of PICAV users use the PICAV system as integration of the public transport system and therefore are flexible (for instance people that visit the pedestrian area and reach it by public transport or are resident in the pedestrian area and need to reach a destination outside the area by public transport).

The second system management strategy
Vehicles can be accessed only at parking lots. PICAVs are assumed to be able to move in a fully automatic way. A fully vehicle based relocation strategy is proposed: vehicles automatically relocate among the parking lots according to the indications of a system supervisor.

A relocation is required when a critical situation occurs. A critical situation occurs when the number of vehicles in a given parking lot at a given time instant goes below the parking lot's low critical threshold. This situation is referred as ZVT, i.e. zero vehicle time (Kek et al. 2009). When this condition occurs, the parking lot has a shortage of vehicles and to avoid a possible queue in the next future a request for a vehicle is generated. The critical threshold could be assumed constant in time or a function of time.

The third system management strategy
In this scenario vehicles are available not only at parking lots but also along the roads. A trip by PICAV may have, as origin or destination, either a parking lot or any position along the roads within the intervention area. When the origin of the user's trip is not a parking lot, a PICAV reaches the user in a fully automatic way. A fully vehicle based relocation strategy is proposed.

Relocations are required:
- When the number of vehicles available at parking lots is below the low critical thresholds (ZVT situation). In this case, the request for a vehicle could be addressed:
a. Firstly to the parking lots where the number of vehicles is above the low buffer threshold (as described in section 1.2.2); among these parking lots, the providing one is selected according to the shortest time or the inventory balancing criteria;
b. Otherwise, to the vehicles parked along the road. In this last case, the nearest vehicle automatically relocates towards the parking lot in shortage.
- When the origin of the user's trip is not a parking lot. In this case, the system supervisor assigns to the user the vehicle nearest to the user position. If the nearest vehicle is in a parking lot, it can be provided only if the number of vehicles available in the parking lot is greater than the low critical threshold of the parking lot.

2.4 The micro simulator
The simulator is written in Python language and it follows an object – oriented logic. The simulator receives in input: the simulation time period, the road network, the transport demand and the car-sharing system characteristics. The simulator allows to track the second-by-second activity of each user, as well as the second-by-second activity of each vehicle. The simulator gives in output the transport system performances, in terms of level of service (LOS) provided to users and in terms of efficiency from the management point of view.

Input data
The input data described in the following are necessary to be able to run all the three system management strategies and to shift from a management scheme to the other, in order to compare their performances. In order to simulate only a given management scheme among the proposed ones, some data may result unnecessary.

The simulation time period

The simulation time period usually starts when the transport system opens and ends when it closes.

The simulation time period could be characterised by peak and off peak phases: for each phase an average pedestrian density k and a PICAV transport demand should be specified.

The road network
The road network includes parking lots, provided with charging stations, and the roads in which PICAV vehicles are allowed to travel. The road network is defined by Open-Street-Map.

Parking lots have been represented through nodes. Each road is divided into sections and each section has been modelized again by a node.

The transport demand
The transport demand is given to the simulator in the form of OD matrixes. OD matrixes are squared and rows/columns refers to nodes. Each cell gives the hourly number of trips by PICAV from the node (origin) the row refers to, to the node (destination) the column refers to.

In the second management strategy, OD matrixes have a number of rows and columns equal to the number of parking lots.

2.5 Output data
Level of Service (LOS)
Level of Service (LOS): LOS measurement is assessed, based on the statistical distribution of waiting times. Castangia and Guala (2011) proposed a scale using as a reference the 50th , 90th and 95th percentile of users waiting times: the scale is provided in following table. All conditions defining a specific LOS should be met.

LOS Waiting time (minutes) not greater than:
50th percentile 90th percentile 95th percentile
A 0.5 1 1.5
B 1 2 3
C 1.5 3 5
D 2.5 5 8
E 4 8 10
F worse worse worse

This table represents LOS measurement based on the statistical distribution of waiting times. (Castangia and Guala, 2011).

Efficiency
From the management point of view, the transport system efficiency is inversely proportional to the PICAV fleet dimension and the number of required relocation trips. Another measure of the system efficiency is the ratio between the number of PICAV vehicles available or occupied by users at each time instant and the PICAV fleet dimension.

2.6 The optimization algorithm
Referring to the first system management strategy, a methodology to optimize the fleet dimension and its distribution among the parking lots has been proposed in Cepolina and Farina (2012). The faced problem was an optimisation problem where the cost function to be minimized takes into account both the transport system cost and the users cost that depend on waiting times. A random search algorithm was adopted. Further details on this optimization procedure are also available in the PICAV report: D4.1 System management strategies.

As it concerns the second and third system management strategies, a methodology to assess the best fleet dimension and the best values for the low critical and low buffer thresholds in order to minimize the users costs (expressed in terms of waiting times) and the system costs (expressed in term of cost of relocation and cost of purchasing the fleet) has been developed. As the objective function was heavily sensitive on the fleet dimension, and slightly sensitive on the thresholds values, in order to obtain a better solution in a shorter amount of time, a parallel optimization of the fleet in one machine, and of the thresholds in the other machine, has been approached. This optimization procedure has been submitted for publication in Cepolina and Farina, submitted for publication to Transportation Research C. Further details on this optimization procedure are also available in the PICAV report: D4.1 System management strategies.

More details about the optimization procedure will be provided herewith.

The constrained minimization problem

The problem of assessing the transport system parameters (i.e. the fleet dimension and the thresholds) is formulated as a constrained minimisation problem. The function z to be minimised is a linear combination of the transport system cost and of the user's costs, subject to the constraint that the level of service results not lower than E. Both costs refer to the daily reference time period and their units are EUROS/day.

The chosen cost is obtained as a function of:
- nv is the fleet dimension and cv is the cost of each vehicle;
- r is the discount rate, equal to 8%;
- lt is the vehicle lifetime;
- cw is the cost of each minute of user waiting time, cr is the cost of each minute of relocation;
- twi is the waiting time of each user in minutes, t rj is the relocation time of each relocation procedure in minutes;
- in the first management strategy, the relocation cost is equal to zero as it is performed directly by the transport system users.

Both the user and the relocation times depend on the solution vector s. This dependence is not analytical: user waiting times and relocation times are determined through the simulator. The solution vector s contains the transport system parameters to be optimized: they are given in input to the simulator, and the simulator provides in output the distribution of user waiting times and the total relocation duration.

The solution vector s is composed of:
- in the first management strategy, of nine components. Each component refers to a parking lot and its value is the number of vehicles at the beginning of the simulation;
- in the second management strategy, of the following three components:

a) the fleet dimension, assumed equally distributed among the various parking lots at the beginning of the simulation;
b) the low critical threshold values, assumed equal at each parking lot and constant during the simulation;
c) the low buffer threshold values, assumed equal at each parking lot and constant during the simulation.
The constraint is the percentile of user waiting time, expressed in minutes.

The Simulated Annealing
A Simulated Annealing algorithm has been chosen to solve the minimization problem.

At each step of the algorithm the vector solution s is updated. The transition rule regards the exploration of the search space: from a given vector solution s, a new vector s n is selected in the neighbourhood of s.

Starting from the vector solution s, the cost function value is assessed through the micro simulator. At each iteration of the Simulated Annealing algorithm, the simulator receives in input the vector solution s and provides in output the cost function.

Parallel optimization
For what regards the second and third management strategy, the overall optimization procedure has shown a serious problem: the objective function is heavily dependent on the fleet dimension (first component of vector s) and slightly dependent on threshold values (second and third components of vector s). This causes some difficulties in the calculation of the optimal solution for what regards the threshold values, which result in the slowdown of the optimization algorithm. Therefore, it has been decided to split the search space into two components: on a processor the objective function is kept dependent only on the fleet dimension and all threshold values are fixed, while on the other processor the objective function depends only on the low critical and buffer thresholds while the fleet dimension is kept constant. Further details on the parallel optimization are described in the report D4.1. System management strategies.

The PICAV simulator website
The PICAV simulator website is written in HTML (with CSS) and Javascript (JS) and has three sections:
- A journey planner;
- A management options interface;
- A display for the waiting time at each parking lot.
These three sections will now be described followed by a description of how the website and the simulator code interact.

Overview of the Journey planner
The journey planner can be used to demonstrate the increased accessibility of a city to users due to the implementation of the PICAV system. It also allows end-users to help plan and optimize the PICAV system by, and this again can help to optimize the pedestrian areas which should be made available to PICAV users. This user-input has been used within the PICAV project with the active involvement of LaCruna, who have helped to develop the street network available to PICAV's. The journey planner incorporates running the simulator as well as the specific journey being tested.

Software used to create the journey planner
OSM OpenStreetMap is a free editable map of the world; it allows people to view, edit and use geographical data in a collaborative way (see http://www.openstreetmap.org/ online). It is possible via the OSM website to add data such as identifying a building as a restaurant or to add routes taken from a GPS device. For the PICAV project the geographical data for the cities in question (Genoa and Barreiro) have been downloaded from the OSM website. This database has then been edited to identify streets as accessible/not accessible to a PICAV by adding the tag PICAV= yes or PICAV= no to the locally stored database. This distinction is important as it mean we have not added the PICAV network to the global OpenStreetMap, however, the data could be added at a later date if we so wished. The roads which were suitable for PICAVs were decided with the help of LaCruna in the case of Genoa and with the aid of TCB in the case of Barreiro.

The journey planner has the following functionality:
- Allows the new user to choose a starting point (origin) and a destination from the map. This can be a point of interest such as a restaurant or just a location.
- Allows the user to set the 'booking time' which is the relative speed with which they can access the vehicle and also the time they wish to undertake the journey.
When the user hits 'Go' two things happen:
- The simulator code is run and the output file containing the waiting times at each of the parking lots at each minute are displayed.
- The individual journey is routed and displayed, showing the walking/wheeling route to the PICAV, the PICAV journey and then the walking/wheeling route to the final destination.
The outputs can be seen by sliding the Simulator timeline. The outputs take two forms. The first displays the number of users waiting at each parking lot at each time iteration (minute) the simulator was running from. The second shows the individual's route.

Website and simulator interaction
The website and simulator code interact in the following ways:
- The website calls the simulator code (simulatore.py) when the 'go' button is pressed via a php script (routing.php);
- The waiting times of users resulting from the simulator code are stored in a comma separated file (csv file) and these are then displayed in the waiting times boxes as the slider is moved;
- The management options of the simulator can be changed in the web interface, which are stored as cookies and are then read by the simulator code.

Management options
The following management options can be changed in the website:
- The management strategy to be tested;
- The start times of the time periods which are defined in the simulator. The end time is assumed to be 1 minute before the proceeding start time);
- The demand level (peak or off-peak) for PICAV in each time period;
- The pedestrian density for each time period;
- The fleet dimensions. This is achieved via selecting the number of PICAVs for each parking lot. The internal parking lots are disabled in the case of no parking lots.

Potential Impact:

The main impact of the PICAV project is the development of an urban transport system ensuring better accessibility for all. The project results improve the outdoor autonomy of weak people and the quality of life of citizens. The new eco-sustainable personal vehicle covers a need and offers a good opportunity to green transport opening a new market branch in the automotive industry. Furthermore, the PICAV system will help reducing the urban decline of the historic city centers by making them easily accessible by everybody, reinforcing the security of people living and frequenting the old city, all of this without introducing noise and pollution due to the zero emissions of the vehicle.

The combination of the prototype and simulator along with journey planner allow authorities to easily test how best to implement a PICAV in their city.

The innovations achieved and demonstrated within the PICAV project, mainly:
- new architecture with wheeled feet inspired from state of the art mobile robots and space exploration rovers;
- 4 driving wheels arranged on 2 wheeled feet;
- Innovative suspensions working on 2 levels: mechanical passive with full active correction (power efficiency);
- Seat designed and sensorized for user comfort and adaptability;
- Guarantee even seat-body contact pressure;
- Driving assistance and Networking;
are the basic features for guaranteeing:
- Social impact by improving the autonomy margins of weak people and improving the citizen quality of life. By helping to reduce the urban decline of historic city centers;
- Economic impact by opening new market branches in automotive industry all along the electric vehicles supply chain.
In more details social impact refers to:
- Reduce the urban decline of the historic city environment by making it easily accessible by everybody; by reinforcing the security of people living and frequenting the old city; all of this without introducing noise and pollution due to emissions;
- Extend green areas without penalize weak mobility users or shopping centres by allowing PICAVing besides walking and cycling.

Dissemination activities
PICAV is a social and industry oriented initiative where ideas and concepts are contributed significantly by the academic partners. This has permitted innovative results, but having a strong aim in the possible industrialization of them.

The dissemination has been carried out and will continue in order to bring forward the innovative vehicle prototype into a commercialized and marketable product that will be appealing for everyone. Different target markets have been recognized; the results have been presented and will be presented continuously to the following markets:
- Automotive industry;
- Mobile robotics and general robotics industry;
- Transport industry;
- Urban mobility responsible;
- Sustainable transport authorities;
- Public administrations.
Dissemination is essential to allow for community generation within the consortium and to close the gap between collaborating communities.

Further important objectives for dissemination and exploitation are:
- To communicate and disseminate the developed technology across European countries and internationally;
- To prepare and pursue the exploitation of the project results.

HMS Rome, September, 12-13-14-2011
(see http://www.msc-les.org/conf/hms2011/ online)
The PICAV transport system simulator was presented in two separate during the session: Innovative Transport Systems for Urban Areas of the 13rd International Conference on Harbor, Maritime & Multimodal Logistics Modelling and Simulation on September 2011. The session focused on a sustainable mobility within urban areas. New vehicles, vehicle fleet management strategies, traveler information services, route guidance were some of the topics this session concerned.

The first Innovation Convention of EC, Brussels, December 05-06-2011
The first generation of the PICAV simulator was demonstrated at the First Innovation Convention of the EC. The graphical user display showing metrics of waiting times, number of PICAVs and number of people waiting at each station was demonstrated and many interesting discussions were had with regards the metrics transport operators would like to see and how best to display these results in a way that real-time decisions about fleet management strategies could be made.

See the article of Maria Koleva, Brussels, 10 December, 2011 http://www.europost.bg/article?id=3326

Transport Research Arena (TRA), Athens, April 23-26 2012.
The developments made to the simulator were demonstrated at TRA. Lively discussions were had following the proposed new integration of a real-time journey planner to interface with the simulator. There are technical challenges in doing this as it relies on inserting journeys from people into the journeys generated through the simulator code. Advances on integrating Open-Street-Map data into the simulator to provide more accurate journey times was also discussed with others involved in route and journey mapping.

Final demonstration, Barreiro, Portugal, September 24-25 2012
The first generation of the PICAV prototype vehicle was presented at the Final Demonstration for the EC as show the following images.

Salone nautico internazionale Genova, Italy, October 16-18-2012
The first generation of the PICAV prototype vehicle was presented at the Salone Nautico Internazionale of Genova as show the following images.
The Genoa International Boat Show is one of the world's premier boat shows, held annually in Genoa, Italy, typically during the month of October. The exhibition is organised by Fiera di Genova SpA and UCINA, Italy's professional association of the yachting industry, and takes place at Fieristico as well as parts of the old town and port of Genoa.
In the 2012 exhibition 900 exhibitors displayed a complete, wide range of products in 4 pavilions and two marinas totalling 7 kilometres of quays (see http://www.genoaboatshow.com/# online).

"Tutti insieme... nelle piazze del sestiere" Terre di Mare event in Genova, Italy, October 20-2012
The first generation of the PICAV prototype vehicle was presented at the Terre di Mare event in the hystorical center of Genova as show the following images. Terredimare.it is the web portal for tourist accessibility in Liguria. The web site is updated daily by the staff of "Coop. Sociale La Cruna" who run the information point "Terre di Mare" in Genoa (Piazza Matteotti 42r).

The "Cooperativa Sociale La Cruna" has as its primary objective the training and employment of people with reduced mobility, the work is in fact considered a right and duty essential in the process of emancipation and social integration. And in fact in the professional skills you recognize the most effective means of promoting integration policy work of disabled people (see http://www.lacruna.com/ online).

The aim of the event was organize activities that will enliven the historical center of Genova: sports for disabled visitors, tastings and workshops open photographic journey "Portraits in motion" etc. (see http://www.terredimare.it/archivio_view.php?ID=8796&lang=it online).

Robotica 2012 Milano, Italy, November 08-09-2012
The first generation of the PICAV prototype vehicle was presented at the Robotica 2012 event in Milan as show the following images.

Robotica 2012 is the only Italian trade fair event in the field of humanoid and service robots, which took place from 7 to 9 November at fieramilano, Rho. It was visited by 6,300 sector professionals, who determined the important success of this fast growing event.

The fourth edition of Robotica had 66 exhibitors, 56 of which Italian and 10 international. As per European Community rules, Robotica can therefore now be called an international fair. Countries represented were United Kingdom, France, Japan, Russia, Switzerland and Poland. This year, the total exhibition area of the event was 4,400 square metres, more than double that of last year (see http://www.robotica.pro/it_rbt/index_rbt.asp online).

Poster presentation and flyer distribution
Some flyers, poster and factsheet were prepared and distributed within different fairs and conferences:

Involvement of robotics associations
The PICAV project was presented and promoted by different ways:
Liaisons with ATA automotive association in Italy and SIRI (Italian association of robotics and automation) have been addressed.

Links with EUROP (The European Robotics Technology Platform), EURON (The European Robotics Research Network) and CLAWAR associations have been established.

Also a link with Electric Vehicles EV of ArtEnergy in Italy was established.

Involvement of IMS and ManuFuture platforms was addressed.

Exploitation of the results
It is very difficult to quantify the economic impact, mainly because PICAV is a really innovative full electric personal vehicle. New market segment in automotive industry could be open in the next future.

- Thanks to the modular approach adopted, the new modules such as the adaptive seating module, intelligent driving interface, new power module, can be commercialised stand-alone and can be integrated into future generations of electric vehicles or other applications reinforcing the competitiveness of the European automotive industry;
- The technology developed will have larger applications, e.g.: - autonomous delivery of parcels based on destination address inside historical centres of cities as well as inside hospitals, campuses, companies; - goods delivery in the shops in the historical centres; - low cost platforms for urban security. These applications are in line with the city of the future and the security concerns, both very actual and sensible issues. The multidisciplinary approach to design knowledge intensive products and services such as PICAV gives advantage to European manufacturers in the markets of the future;
- The commercial capability of shops in the old cities will be improved, so making the old cities more 'alive', by allowing for the pollution-free delivery of supplies and customers; the PICAV new personal low-cost eco-sustainable vehicle will allow the European industries, included SMEs, to gain competitiveness in the market of electric and intelligent vehicles.

Direct contact with potential users
5th August 2012 – TechWave – Olympics related Technology Innovation Showcase, Ravenscourt, London, August 2012
The PICAV project was chosen to represent UCL research innovations at the TechWave – Olympics related Technology Innovation Showcase. This event took place during the Olympics next to the O2 Centre where the Athletics was taking place. A number of innovation experts (a mixture of entrepreneurs and city administrators), many of whom were on a 1 week visit which was funded by the Mayor of London's office attended the event. The PICAV vehicle and the simulator were presented and two interesting lines of enquiry developed. The first was on the use of cutting edge web technologies to enhance the user experience and the second with Greenwich Council about the implementation of a trial system in Borough of Greenwich, London.

PICAV final meeting, Barreiro, September 2012
The final release of the simulator has been presented at this event. Among the attendees, specially people from the public transport company of Barreiro (TCB) and people from the municipality of Lisbon showed interest in the presented simulator. Two separate presentations were given, the first by Elvezia Cepolina focused on the algorithms driving the simulator; the second by Nick Tyler focused on the development of the GUI with integrated journey planner.

List of Websites:
http://www.picav.
http://www.youtube.com/user/PMARrobotics?feature=g-all-uw

Related documents