CORDIS

DENSE is developing an integrated sensor suite capable of operating in adverse weather. The project focuses on overcoming the problem of limited perception in restricted visibility conditions where accident risks are highest. The novel concept will enhance the sensor functional performance for all weather scenarios and increase the trust associated with the operation of vehicles. The sensor suite includes a novel multipurpose radar concept, an automotive LIDAR working in the short-wave infrared (SWIR) and a SWIR-camera suitable for automotive application. Without a highly reliable sensing system, automated driving in levels 4 and 5 will never be realised. The DENSE sensor suite will enable the higher driving automation levels under more environmental conditions.

The core activities of the project are to define requirements, specify, develop and validate a fused sensor suite with the best cost-benefit ratio, so that a broad market penetration of the novel all-weather sensor system can eventually be achieved.

The overall objectives are:
• To develop and demonstrate an all-weather sensor suite for traffic services, driver assistance and autonomous driving.
• Define functional requirements, specify, develop, test and validate a fused sensor suite with the best cost-benefit ratio.
• Promote the DENSE all-weather sensor system to facilitate and speed up the market introduction.

Key benefits for:
• EU citizens, through a radical improvement of transport safety and a new quality of life, i.e. for the ageing population.
• European economy, through new job and employment opportunities, bridging traditional engineering, software development and environment perception technologies.
• European industry, stimulating environment perception sciences, sensor and IT sectors, providing a large potential for innovation.
• Public finances, by curbing the energy spending through making vehicle and traffic management systems more cost effective.

"For effective management and internal communication, activities in WP1 included online meetings, task distribution and tracking of WP progress with an internal reporting. Activities of WP2 (requirements definition, use cases, benchmark definition) and WP3 (baseline system architecture, test definition and verification concept) were concluded during the first reporting period. Partners in these WPs supported the work of WP4, WP5 and WP6 where major progress was made in P2.

WP4 work in P1 focused on assembling the first prototype sensor working at 1500nm. In P2, WP4 (Sensor Development) activities concentrated on: (a) SWIR camera: chip architecture design (ROIC) and hardware testing. The new design for an advanced higher power 1.55μm laser diode is completed and should allow beyond state-of-the art pulse power. A simulation model was finished and optics design and mechanics designs are ready for manufacturing. // (b) Lidar prototype: detector and electronics were designed, assembled and a prototype was tested in lab conditions at CEREMA with promising results supporting to move from 905 towards 1550. The housing design, assembly of detector, illuminator, mems mirror and sensor control is work in progress. // (c) Radar: simulations ongoing from different multi-static radar concepts. Preliminary results show improved resolution under certain conditions. // (d) Road friction: Concept 1 prototype is working under lab conditions. Feedback from the prototype 1 tests is taken into account in development of the concept 2 prototype.

In P1, WP5 focused on applying and validating the synchronised data output from different sensor types and on the development of in-vehicle data fusion platform. In P2, WP5 focused on documentation and evaluation of the sensor fusion approach. New key components (three new sensors and a novel processing unit consisting of deep neural networks (DNN)) for the sensor suite were tested. The DNNs enhance and fuse sensor signals to an enriched and enhanced multi spectral image. The first sensor fusion tests look promising and showed the advantage of employing a diverse mix of sensors for a better perceptions of the vehicles surroundings. Training data for the DNNs was collected during bad weather season.

In WP6: For evaluation and benchmarking, data of commercial LiDARs and radars was recorded in the Cerema fog chamber as well as in snowy conditions. External communication was continued with regular updates on the website and via twitter. To increase traffic a special interest topic for expert target groups was defined: ""Automated driving in bad weather"". Furthermore, partners presented first results during events: (a) European ITS Congress in Strasbourg, (b) EUCAR reception, (c) scientific day “Smart mobility” organized by the French scientific cluster LABEX ImobS3 and Viameca innovative cluster. A scientific paper was submitted to 21st ITSC 2018, entitled “Automotive LiDAR performance verification in fog and rain”. On request of the reviewers an exploitation plan draft as well as the Validation Plan (D6.3) was submitted."

The main aim of DENSE is to improve the perception capabilities of sensors in all kinds of adverse weather and visibility conditions. DENSE strives to make a difference through both a) safely handling the complex task of autonomy by merging three different technologies and b) developing an all-weather sensor suite at a fraction of the cost. This will bring Europe to the forefront of the sensor market and DENSE partners will be in the position to offer a technology solution directly integrated to existing environmental perception systems.

During P1, the consortium enhanced knowledge sharing. In P2, first technical results were achieved:
SWIR gated cam: ROIC design was finalised with total power between 1W and 1.2W.
Laser: Three different designs approaches were fabricated for performance comparison (AIGaInAs reference 4QW, ALGaInAs LOC 4QW and InGaAsP LOC with 3QW).
Radar: data acquisition software was developed. Simulations were performed in order to achieve resolution from different multi-static radar concepts. First measurements show angular resolution improvements with two radars depending on location of the targets. Angular resolution improvements are ~5.5° (=16.9°-11.4°) when the distance between radar 1. 1 and radar 1.2 is 0.62 m. When the distance is increased to 1.28 m, the angular resolution is ~6.7° (=16.9°-10.2°).
Road friction sensor: The first prototype is available and installed in 4 passenger cars and in several heavy-duty vehicles.

The success of DENSE in achieving its objectives will be indicated, among others, by the number of stakeholders that engage in discussions and in the verification of the sensor suite. As the project activities progress we will better understand how system components interact with each other, vehicles and drivers and how they are enabled or constrained. The efforts are driven primarily by the need to make mobility accident-free with new services in place, whilst guaranteeing cost-efficient use of all the resources in play. We expect to demonstrate that the DENSE all-weather sensor suite can be extremely useful to completely understand all parameters of the driving scene ahead when visibility is restricted.