TY - GEN
T1 - Safe Reinforcement Learning using Data-Driven Predictive Control
AU - Selim, Mahmoud
AU - Alanwar, Amr
AU - Watheq El-Kharashi, M.
AU - Abbas, Hazem M.
AU - Johansson, Karl H.
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Reinforcement learning (RL) algorithms can achieve state-of-the-art performance in decision-making and continuous control tasks. However, applying RL algorithms on safety-critical systems still needs to be well justified due to the exploration nature of many RL algorithms, especially when the model of the robot and the environment are unknown. To address this challenge, we propose a data-driven safety layer that acts as a filter for unsafe actions. The safety layer uses a data-driven predictive controller to enforce safety guarantees for RL policies during training and after deployment. The RL agent proposes an action that is verified by computing the data-driven reachability analysis. If there is an intersection between the reachable set of the robot using the proposed action, we call the data-driven predictive controller to find the closest safe action to the proposed unsafe action. The safety layer penalizes the RL agent if the proposed action is unsafe and replaces it with the closest safe one. In the simulation, we show that our method outperforms state-of-the-art safe RL methods on the robotics navigation problem for a Turtlebot 3 in Gazebo and a quadrotor in Unreal Engine 4 (UE4).
AB - Reinforcement learning (RL) algorithms can achieve state-of-the-art performance in decision-making and continuous control tasks. However, applying RL algorithms on safety-critical systems still needs to be well justified due to the exploration nature of many RL algorithms, especially when the model of the robot and the environment are unknown. To address this challenge, we propose a data-driven safety layer that acts as a filter for unsafe actions. The safety layer uses a data-driven predictive controller to enforce safety guarantees for RL policies during training and after deployment. The RL agent proposes an action that is verified by computing the data-driven reachability analysis. If there is an intersection between the reachable set of the robot using the proposed action, we call the data-driven predictive controller to find the closest safe action to the proposed unsafe action. The safety layer penalizes the RL agent if the proposed action is unsafe and replaces it with the closest safe one. In the simulation, we show that our method outperforms state-of-the-art safe RL methods on the robotics navigation problem for a Turtlebot 3 in Gazebo and a quadrotor in Unreal Engine 4 (UE4).
KW - Reinforcement learning
KW - robot safety
KW - task and motion planning
UR - http://www.scopus.com/inward/record.url?scp=85147551048&partnerID=8YFLogxK
U2 - 10.1109/ICCSPA55860.2022.10018994
DO - 10.1109/ICCSPA55860.2022.10018994
M3 - Conference contribution
AN - SCOPUS:85147551048
T3 - 2022 5th International Conference on Communications, Signal Processing, and their Applications, ICCSPA 2022
BT - 2022 5th International Conference on Communications, Signal Processing, and their Applications, ICCSPA 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 5th International Conference on Communications, Signal Processing, and their Applications, ICCSPA 2022
Y2 - 27 December 2022 through 29 December 2022
ER -