A real-world application of Markov chain Monte Carlo method for Bayesian trajectory control of a robotic manipulator
| Autor: | Arda Agababaoglu, Sinan Yildirim, Ahmet Onat, Vahid Tavakol Aghaei |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
0209 industrial biotechnology
Mathematical optimization Computer science Bayesian probability TJ Mechanical engineering and machinery 02 engineering and technology Bayesian inference Field (computer science) symbols.namesake 020901 industrial engineering & automation 0202 electrical engineering electronic engineering information engineering Reinforcement learning Electrical and Electronic Engineering Instrumentation business.industry Applied Mathematics 020208 electrical & electronic engineering Markov chain Monte Carlo Robotics Computer Science Applications Control and Systems Engineering symbols Artificial intelligence business Intelligent control Robotic arm |
| DOI: | 10.1016/j.isatra.2021.06.010 |
| Popis: | Reinforcement learning methods are being applied to control problems in robotics domain. These algorithms are well suited for dealing with the continuous large scale state spaces in robotics field. Even though policy search methods related to stochastic gradient optimization algorithms have become a successful candidate for coping with challenging robotics and control problems in recent years, they may become unstable when abrupt variations occur in gradient computations. Moreover, they may end up with a locally optimal solution. To avoid these disadvantages, a Markov chain Monte Carlo (MCMC) algorithm for policy learning under the RL configuration is proposed. The policy space is explored in a non-contiguous manner such that higher reward regions have a higher probability of being visited. The proposed algorithm is applied in a risk-sensitive setting where the reward structure is multiplicative. Our method has the advantages of being model-free and gradient-free, as well as being suitable for real-world implementation. The merits of the proposed algorithm are shown with experimental evaluations on a 2-Degree of Freedom robot arm. The experiments demonstrate that it can perform a thorough policy space search while maintaining adequate control performance and can learn a complex trajectory control task within a small finite number of iteration steps. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|