WHY OFF-THE-SHELF PHYSICS SIMULATORS FAIL IN EVALUATING FEEDBACK CONTROLLER PERFORMANCE - A CASE STUDY FOR QUADRUPEDAL ROBOTS
M. NEUNERT, Thiago Boaventura, J. BUCHLI
- 发表年份
- 2016
- 引用次数
- 11
摘要
Simulators are essential to robotics research since they help us to understand physical properties of the robotic system as well as test control and motion planning approaches in a quick and safe manner.Yet most robotics simulators rely entirely on physics engines and do not model actuator dynamics, noise or delay.However, as we demonstrate in this work, modeling these effects is essential when evaluating the performance of closed loop controllers.In fact, neglecting them is one reason why controllers do not transfer well from simulation to hardware.While the presented study applies to most multi-link robots, our parameter choice is inspired by state-of-the-art quadrupedal robots.
关键词
相关论文
Statistical Learning Theory
Yuhai Wu, Vladimir Vapnik
1999
Artificial intelligence: a modern approach
1995
Applied Nonlinear Control
Jean-Jacques Slotine, Weiping Li
1991
A new optimizer using particle swarm theory
R.C. Eberhart, James Kennedy
2002