Timezone: »

 
Poster
Constrained Update Projection Approach to Safe Policy Optimization
Long Yang · Jiaming Ji · Juntao Dai · Linrui Zhang · Binbin Zhou · Pengfei Li · Yaodong Yang · Gang Pan

@

Safe reinforcement learning (RL) studies problems where an intelligent agent has to not only maximize reward but also avoid exploring unsafe areas. In this study, we propose CUP, a novel policy optimization method based on Constrained Update Projection framework that enjoys rigorous safety guarantee. Central to our CUP development is the newly proposed surrogate functions along with the performance bound. Compared to previous safe reinforcement learning meth- ods, CUP enjoys the benefits of 1) CUP generalizes the surrogate functions to generalized advantage estimator (GAE), leading to strong empirical performance. 2) CUP unifies performance bounds, providing a better understanding and in- terpretability for some existing algorithms; 3) CUP provides a non-convex im- plementation via only first-order optimizers, which does not require any strong approximation on the convexity of the objectives. To validate our CUP method, we compared CUP against a comprehensive list of safe RL baselines on a wide range of tasks. Experiments show the effectiveness of CUP both in terms of reward and safety constraint satisfaction. We have opened the source code of CUP at https://github.com/zmsn-2077/CUP-safe-rl.

Keywords: Reinforcement Learning · Deep Reinforcement Learning · Constrained MDP

Author Information

Long Yang (Zhejiang University)
Jiaming Ji
Juntao Dai (Zhejiang University)
Linrui Zhang (Tsinghua University)
Binbin Zhou (Zhejiang University City College)
Pengfei Li (Zhejiang University)
Yaodong Yang (AIG)
Gang Pan (Zhejiang University)

More from the Same Authors