Distributed Deep Reinforcement Learning (DRL) aims to leverage more computational resources to train autonomous agents with less training time. Despite recent progress in the field, reproducibility issues have not been sufficiently explored. This paper first shows that the typical actor-learner framework can have reproducibility issues even if hyperparameters are controlled. We then introduce Cleanba, a new open-source platform for distributed DRL that proposes a highly reproducible architecture. Cleanba implements highly optimized distributed variants of PPO and IMPALA. Our Atari experiments show that these variants can obtain equivalent or higher scores than strong IMPALA baselines in moolib and torchbeast and PPO baseline in CleanRL. However, Cleanba variants present 1) shorter training time and 2) more reproducible learning curves in different hardware settings.

There has been significant progress in developing reinforcement learning (RL) training systems. Past works such as IMPALA, Apex, Seed RL, Sample Factory, and others, aim to improve the system's overall throughput. In this paper, we aim to address a common bottleneck in the RL training system, i.e., parallel environment execution, which is often the slowest part of the whole system but receives little attention. With a curated design for paralleling RL environments, we have improved the RL environment simulation speed across different hardware setups, ranging from a laptop and a modest workstation, to a high-end machine such as NVIDIA DGX-A100. On a high-end machine, EnvPool achieves one million frames per second for the environment execution on Atari environments and three million frames per second on MuJoCo environments. When running EnvPool on a laptop, the speed is 2.8x that of the Python subprocess. Moreover, great compatibility with existing RL training libraries has been demonstrated in the open-sourced community, including CleanRL, rl_games, DeepMind Acme, etc. Finally, EnvPool allows researchers to iterate their ideas at a much faster pace and has great potential to become the de facto RL environment execution engine. Example runs show that it only takes five minutes to train agents to play Atari Pong and MuJoCo Ant on a laptop. EnvPool is open-sourced at https://github.com/sail-sg/envpool.

In this paper, we present Tianshou, a highly modularized Python library for deep reinforcement learning (DRL) that uses PyTorch as its backend. Tianshou intends to be research-friendly by providing a flexible and reliable infrastructure of DRL algorithms. It supports online and offline training with more than 20 classic algorithms through a unified interface. To facilitate related research and prove Tianshou's reliability, we have released Tianshou's benchmark of MuJoCo environments, covering eight classic algorithms with state-of-the-art performance. We open-sourced Tianshou at https://github.com/thu-ml/tianshou/.

In high-throughput drug screening applications, as manual drug sample delivery is time-consuming and error-prone, there is an urgent need for accurate and efficient drug sample delivery biochip for large-scale microwell arrays. This paper proposes a new microfluidic biochip architecture, where drugs are automatically prepared with different concentration values, and then delivered into multiple microwells. For large-scale drug sample delivery biochips, the routing of drug sample delivery channels is a very challenging task without effective routing solutions. This paper proposes an ultrafast rule-based escape routing method, called URBER, to address the large-scale routing of drug sample delivery channels, which scales well in both runtime and memory even for a very large problem size. URBER runs very fast because it routes channels based on a set of predefined rules, which avoids runtime consumed in solution space exploration. All benchmarks for 30 ≤ N, M ≤ 100 have been tested, where N and M are the number of columns and rows of the terminal array. Among these benchmarks, about ~91.9% are routed with optimal solutions, and the runtime is order of magnitudes faster than optimal min-cost flow-based methods (speedup is from ~600 to ~340 k). Specifically, for all benchmarks with (M/N) E ((3/4), (4/3)), optimal routing solutions are always obtained. URBER also shows promise of routing large-scale designs with up to 500 k terminals efficiently.

Learning rational behaviors in First-person-shooter (FPS) games is a challenging task for Reinforcement Learning (RL) with the primary difficulties of huge action space and insufficient exploration. To address this, we propose a hierarchical agent based on combined options with intrinsic rewards to drive exploration. Specifically, we present a hierarchical model that works in a manager-worker fashion over two levels of hierarchy. The high-level manager learns a policy over options, and the low-level workers, motivated by intrinsic reward, learn to execute the options. Performance is further improved with environmental signals appropriately harnessed. Extensive experiments demonstrate that our trained bot significantly outperforms the alternative RL-based models on FPS games requiring maze solving and combat skills, etc. Notably, we achieved first place in VDAIC 2018 Track(1).