Periodic Reporting for period 2 - GRIFFIN (General compliant aerial Robotic manipulation system Integrating Fixed and Flapping wings to INcrease range and safety)
Periodo di rendicontazione: 2020-05-01 al 2021-10-31
The goal of GRIFFIN is the development of a new generation of bioinspired aerial robots with manipulation capabilities that will be able to safely operate in sites where rotorcrafts cannot do it and physically interact with people.
In GRIFFIN we are deriving a unified framework with methods, tools and technologies for the development of robots that can fly minimizing energy consumption by the combination of gliding and flapping modes of flight, perch on constrained surfaces, fold the wings and perform dexterous manipulation.
The flapping unmanned aerial vehicles, also called ornithopters, that we propose, improve the existent aerial manipulators based on multirotor platforms in terms of longer flight, duration of missions and safety in proximity to humans. The absence of propellers makes such vehicles more resistant to physical contact, permitting flight in cluttered environments, and to collaborate with humans. Importantly, the provision of thousands of species of birds that have already mastered the challenging task of flapping flight is a rich source of solutions.
In GRIFFIN we have developed prototypes that are able to carry a payload of 100% of its own weigh including all the electronics, computer and sensors for autonomous operation. We also developed morphing wings, bioinspired tails and claws. These prototypes are being produced with additive manufacturing integrating parts with different materials.
Moreover, we have developed kinematic, dynamic and aerodynamic models, as well as adaptive control and guidance techniques, that have been validated both indoors and outdoors.
GRIFFIN has produced the first fully autonomous ornithopters with embedded environment perception, control and decision capabilities that have been validated indoors and outdoors. Particularly, we have designed, developed and validated prototypes, with event-cameras and conventional cameras, providing images that are processed on-board to build maps of the environment, estimate the position of the ornithopter, detect perching sites and perform perching manoeuvres on a small surface.
The manipulation can be performed by maintaining fixed contact with a surface, such as a pole or a pipe, by means of one or more limbs, and manipulating with others, overcoming the current limitations of accurate manipulation in free flying. Particularly, we have developed prototypes that have demonstrated manipulation capabilities by maintaining the equilibrium of forces when perched and following a desired path with the end-effector. This allows several possible applications, such as contact inspection with an ultrasonic sensor mounted in the end-effector.
New software tools are being developed to facilitate the design and implementation of the above complex robotic systems. Thus, configurations with different complexity could be derived depending on the requirements of flight endurance and manipulation tasks, from simple grasping to more complex dexterous manipulation.
The GRIFFIN bioinspired aerial robots could be applied in a large variety of tasks, such as the transportation and delivery of small loads, the contact inspection in sites difficult to reach, and even the assistance in search and rescue operations being able to measure biometric parameters of victims.
In GRIFFIN we are deriving a unified framework with methods, tools and technologies for the development of robots that can fly minimizing energy consumption by the combination of gliding and flapping modes of flight, perch on constrained surfaces, fold the wings and perform dexterous manipulation.
The flapping unmanned aerial vehicles, also called ornithopters, that we propose, improve the existent aerial manipulators based on multirotor platforms in terms of longer flight, duration of missions and safety in proximity to humans. The absence of propellers makes such vehicles more resistant to physical contact, permitting flight in cluttered environments, and to collaborate with humans. Importantly, the provision of thousands of species of birds that have already mastered the challenging task of flapping flight is a rich source of solutions.
In GRIFFIN we have developed prototypes that are able to carry a payload of 100% of its own weigh including all the electronics, computer and sensors for autonomous operation. We also developed morphing wings, bioinspired tails and claws. These prototypes are being produced with additive manufacturing integrating parts with different materials.
Moreover, we have developed kinematic, dynamic and aerodynamic models, as well as adaptive control and guidance techniques, that have been validated both indoors and outdoors.
GRIFFIN has produced the first fully autonomous ornithopters with embedded environment perception, control and decision capabilities that have been validated indoors and outdoors. Particularly, we have designed, developed and validated prototypes, with event-cameras and conventional cameras, providing images that are processed on-board to build maps of the environment, estimate the position of the ornithopter, detect perching sites and perform perching manoeuvres on a small surface.
The manipulation can be performed by maintaining fixed contact with a surface, such as a pole or a pipe, by means of one or more limbs, and manipulating with others, overcoming the current limitations of accurate manipulation in free flying. Particularly, we have developed prototypes that have demonstrated manipulation capabilities by maintaining the equilibrium of forces when perched and following a desired path with the end-effector. This allows several possible applications, such as contact inspection with an ultrasonic sensor mounted in the end-effector.
New software tools are being developed to facilitate the design and implementation of the above complex robotic systems. Thus, configurations with different complexity could be derived depending on the requirements of flight endurance and manipulation tasks, from simple grasping to more complex dexterous manipulation.
The GRIFFIN bioinspired aerial robots could be applied in a large variety of tasks, such as the transportation and delivery of small loads, the contact inspection in sites difficult to reach, and even the assistance in search and rescue operations being able to measure biometric parameters of victims.
We have implemented the action following the Work-Programme. The major achievements are:
• Modelling. Ornithopter models combining flapping and gliding. The models integrate longitudinal dynamics and aerodynamics with functions for lift and thrust of flapping wings according to potential theory. Validation with prototypes in the indoor testbed and by outdoor experiments.
• Control. Adaptive nonlinear control systems, mainly focused on the autonomous perching manoeuver, which are able to adapt the unknown aerodynamic coefficients during the flight. The controllers have been integrated in the prototypes and validated in the indoor testbed.
• Perception. Focused on the development of novel methods that consider the nature of the flapping-wing flight. Implementation of event-based vision, with adaptation of asynchronous processing, demonstrating detection and tracking under different illumination conditions, aggressive manoeuvring and motion blur. Visual servoing techniques for autonomous perching have also being researched.
• Manipulation. Methods performing manipulation while perching by maintaining the equilibrium of forces. Several control techniques were developed and tested in a prototype which mimics the anatomy of birds. Compliant dual arm manipulation with a tail has been also implemented.
• Tools: Computational Fluid Dynamic for aerodynamic design and simulation tools for bird landing.
• Ornithopter prototypes. Three Proof of Concept Prototypes (PCP) have been developed by using additive manufactured mechanical components and integrating customised lightweight autopilot and computational capabilities. Research on morphing parts of the flapping wing system is being developed. Macro fibers composites have been applied for morphing a bio-inspired tail.
• Manipulation prototypes. Three small scale dual arm prototypes (250 mm reach, ~0.2 kg weight) have been developed using micro metal gear motors and high performance harmonic drive actuators. Moreover, bioinspired actuators of joints and end effectors are being developed. Twisted and coiled polymers (TCP) and shape memory alloys (SMA) have been used for closing end-effectors.
• Visual perching preliminary testing with multi-rotors. The work includes machine learning and convolutional neural network. Furthermore, soft landing gear systems capable of adjusting to their shape and diameter have been developed. The system has been tested both indoors and outdoors.
More information of the project including Dissemination (Publications and Project Presentations), Media Content and News are in the Web-site of the project: https://griffin-erc-advanced-grant.eu/
• Modelling. Ornithopter models combining flapping and gliding. The models integrate longitudinal dynamics and aerodynamics with functions for lift and thrust of flapping wings according to potential theory. Validation with prototypes in the indoor testbed and by outdoor experiments.
• Control. Adaptive nonlinear control systems, mainly focused on the autonomous perching manoeuver, which are able to adapt the unknown aerodynamic coefficients during the flight. The controllers have been integrated in the prototypes and validated in the indoor testbed.
• Perception. Focused on the development of novel methods that consider the nature of the flapping-wing flight. Implementation of event-based vision, with adaptation of asynchronous processing, demonstrating detection and tracking under different illumination conditions, aggressive manoeuvring and motion blur. Visual servoing techniques for autonomous perching have also being researched.
• Manipulation. Methods performing manipulation while perching by maintaining the equilibrium of forces. Several control techniques were developed and tested in a prototype which mimics the anatomy of birds. Compliant dual arm manipulation with a tail has been also implemented.
• Tools: Computational Fluid Dynamic for aerodynamic design and simulation tools for bird landing.
• Ornithopter prototypes. Three Proof of Concept Prototypes (PCP) have been developed by using additive manufactured mechanical components and integrating customised lightweight autopilot and computational capabilities. Research on morphing parts of the flapping wing system is being developed. Macro fibers composites have been applied for morphing a bio-inspired tail.
• Manipulation prototypes. Three small scale dual arm prototypes (250 mm reach, ~0.2 kg weight) have been developed using micro metal gear motors and high performance harmonic drive actuators. Moreover, bioinspired actuators of joints and end effectors are being developed. Twisted and coiled polymers (TCP) and shape memory alloys (SMA) have been used for closing end-effectors.
• Visual perching preliminary testing with multi-rotors. The work includes machine learning and convolutional neural network. Furthermore, soft landing gear systems capable of adjusting to their shape and diameter have been developed. The system has been tested both indoors and outdoors.
More information of the project including Dissemination (Publications and Project Presentations), Media Content and News are in the Web-site of the project: https://griffin-erc-advanced-grant.eu/
GRIFFIN is working on the development of a new generation of autonomous bioinspired aerial robots capable of physically interacting with their environment.
We are developing new ornithopters that combine flapping with gliding to save energy. New modeling and control methods are being developed and tested for these ornithopters. We are also developing autonomous perception techniques. In particular, vision systems based on event cameras are being developed. These techniques allow the implementation of smart behaviors, including the fully autonomous perching.
In parallel, we are developing manipulation capabilities that will be performed while perching maintaining balance. We have developed a new control technique for flapping aerial robots being aware of the hard constraints imposed by aerial robots while perched.
Furthermore, we are developing tools to be used in the development of new ornithopters. These include Computational Fluid Dynamic (CFD) for aerodynamic design and simulation tools for flying and landing.
All of the above methods and technologies are being demonstrated by prototypes. We are developing new bio-inspired parts and actuators. In particular, the technologies for tail and claw morphing have already been developed and tested. Also embedded control systems have been developed and tested.
In the first period we have built three ornithopter prototypes and different manipulation systems.
In the coming periods, we have to fully develop all the methods and technologies mentioned above, including modeling, control, perception and planning. New technologies for bioinspired actuators will also be developed, including wing folding.
Significant efforts should also be devoted to integrating methods and technologies into new systems and prototypes. In particular, GRIFFIN prototypes will be developed integrating manipulation capabilities in the ornithopter.
Finally, we are also pursuing the development of demonstrators in relevant applications, including interaction with people and the transportation of lightweight parts and samples.
We are developing new ornithopters that combine flapping with gliding to save energy. New modeling and control methods are being developed and tested for these ornithopters. We are also developing autonomous perception techniques. In particular, vision systems based on event cameras are being developed. These techniques allow the implementation of smart behaviors, including the fully autonomous perching.
In parallel, we are developing manipulation capabilities that will be performed while perching maintaining balance. We have developed a new control technique for flapping aerial robots being aware of the hard constraints imposed by aerial robots while perched.
Furthermore, we are developing tools to be used in the development of new ornithopters. These include Computational Fluid Dynamic (CFD) for aerodynamic design and simulation tools for flying and landing.
All of the above methods and technologies are being demonstrated by prototypes. We are developing new bio-inspired parts and actuators. In particular, the technologies for tail and claw morphing have already been developed and tested. Also embedded control systems have been developed and tested.
In the first period we have built three ornithopter prototypes and different manipulation systems.
In the coming periods, we have to fully develop all the methods and technologies mentioned above, including modeling, control, perception and planning. New technologies for bioinspired actuators will also be developed, including wing folding.
Significant efforts should also be devoted to integrating methods and technologies into new systems and prototypes. In particular, GRIFFIN prototypes will be developed integrating manipulation capabilities in the ornithopter.
Finally, we are also pursuing the development of demonstrators in relevant applications, including interaction with people and the transportation of lightweight parts and samples.