Demos of research activities of MRS group members

Quick links: Motion planning demos, Modular robotics demos

Motion planning

Example of solution of 3D Bugtrap benchmark probem, where the task is to find path for a stick robot. The problem is difficult due to narrow passage. The video shows a result obtained using RRT-IS (RRT with Iterative scaling): V Vonasek, J Faigl, T Krajnik and L Preucil. A Sampling Schema for Rapidly Exploring Random Trees Using a Guiding Path. In Proceedings of the 5th European Conference on Mobile Robots. 2011, 201–206. See Motion planning page for details.

Example of solution of 3D Hedge in the cage problem, where the task is to remove the spiky robot out of the cage. The cage has five identical windows with width similar to the radius of the robot's sphere. The video shows a result obtained using RRT-IS (RRT with Iterative scaling): V Vonasek, J Faigl, T Krajnik and L Preucil. A Sampling Schema for Rapidly Exploring Random Trees Using a Guiding Path. In Proceedings of the 5th European Conference on Mobile Robots. 2011, 201–206. See Motion planning page for details.

Example of solution of Alpha puzzle benchmark using RRT with RRT-IS: V Vonasek, J Faigl, T Krajnik and L Preucil. A Sampling Schema for Rapidly Exploring Random Trees Using a Guiding Path. In Proceedings of the 5th European Conference on Mobile Robots. 2011, 201–206. See Motion planning page for details.

And of course: the well known Piano Mover's problem!

Modular robotics

Example of motion planning for a modular robot made of five modules. Locomotion of the robot is generated using Central Pattern Generators. The robot is equipped with four CPG-based motion primitives. A sequence of the primitives is found using RRT-MP algorithm (see Modular robotics page for details).

Example of gait optimization for modular robot. The red module is broken (it cannot be controlled). The optimization starts in a simulation, where four best solutions are selected and further optimized on the HW robot. The second part of the video (00:10) shows the progress of the optimization on real robot. The Fitness value shows actual quality of motion (traveled distances in cm in 30 seconds). The final gait has speed 27 cm/30s = 54 cm/min and it was achieved after 2 minutes of the HW experiment. More info is in the paper: Vonasek, Vojtech; Neumann, Sergej; Oertel, David; Worn, Heinz, "Online motion planning for failure recovery of modular robotic systems," Robotics and Automation (ICRA), 2015 IEEE International Conference on , vol., no., pp.1905,1910, 26-30 May 2015. (see Modular robotics page for details).

A caterpillar-like robot made of five CoSMO modules. The red modules is broken and cannot be controlled. The task is to find motion 'move-left' using a Central Pattern Generator. Optimization is realized using Particle Swarm Optimization.More info is in the paper: Vonasek, Vojtech; Neumann, Sergej; Oertel, David; Worn, Heinz, "Online motion planning for failure recovery of modular robotic systems," Robotics and Automation (ICRA), 2015 IEEE International Conference on , vol., no., pp.1905,1910, 26-30 May 2015. (see Modular robotics page for details).