Space travel with human astronauts is associated with enormous costs and hazards without offering an immediate return on investment. Thus, the exploration of planets is and will primarily be carried out with the help of robotic systems. The robots currently used for the exploration of the Moon and Mars are traditionally rovers (Sojourner, Opportunity, Spirit, Curiosity, Zhurong, Lunokhod 1&2, Yutu) with other traversal methods only recently explored with the Ingenuity helicopter on Mars. Aerial locomotion is extremely difficult on other terrestrial bodies due to either lack of atmosphere (moon, asteroids, comets) or due to a very thin Atmosphere (Mars). Legged locomotion offers traversal advantages due to the inherent versatility of legged systems for locomotion. Thus, future robotic exploratory missions would benefit from legged robotic systems which can traverse difficult terrains and help in carrying out greater scientific work. For robotic systems, exploration of unknown terrain does pose a tremendous challenge since it is impossible to predict all the situations and conditions that could potentially occur, especially when robots are predestined to explore steep lunar/mars craters, which requires great physical abilities and resilience. In this context, this project aims at providing an existing dynamic quadruped robot with enhanced capabilities for on-site modeling and adapting the robot’s system dynamics to cope with unknown gravity conditions and friction as well as dealing with wear and tear using a mixture of analytical modeling techniques and data-driven methods. The results are to be shown in a demanding task at the end of the project in which a quadruped robot performs robust walking on slopes, running, and jumping behaviors over obstacles and backflips not only on Earth gravity conditions but also under low-gravity conditions after model self-adaption. This work aims to push the limits of the physical performance of robots as well as self-adaption capabilities by developing a software control system that could be used as a module to equip future robotics systems with the capacity to cope with the tremendous challenges that a lunar crater exploration would pose on them.
This project aims to achieve the following scientific and technological goals:
- Develop an adaptive real-time control system for quadruped locomotion which accomplishes the following tasks: Adapt the robot parameters (such as joint friction/damping, link inertia, joint faults) online, Estimate and adapt online the environment parameters (such as soil properties, friction),Use the newly adapted parameters to explore and push the limits of the action envelope of the robot.
- Develop a real-time software framework that enables adaptive control described above.
- Demonstrate these capabilities using highly athletic motion (such as jumps, running with flight phases, and backflips) in different environments.
