- In this repository, I find an efficient way to encode crowd behavior information
- I adapt the baseline from CrowdNav
- Refer to my simple baseline with Q-learning, easy to use Robot navagation in crow(updating)
- 14/05/2024: I use attention mechanism from SARL to re-rank human's observable vectors, which human have highest attention scores will feed into last cell of LSTM, the modification reduced average time to goal value from 12.38 to 11.44. Beside that, I test model which trained with circle crossing setting with square crossing setting (cross domain).
- 12/05/2024: Upload baseline with LSTM
- The same with original repository CrowdNav
python train.py --policy lstm_att
- [1] https://github.com/vita-epfl/CrowdNav
- [2] Chen, Yu Fan, et al. "Decentralized non-communicating multiagent collision avoidance with deep reinforcement learning." 2017 IEEE international conference on robotics and automation (ICRA). IEEE, 2017.
- [3] Everett, Michael, Yu Fan Chen, and Jonathan P. How. "Motion planning among dynamic, decision-making agents with deep reinforcement learning." 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2018.
- [4] Chen, Changan, et al. "Crowd-robot interaction: Crowd-aware robot navigation with attention-based deep reinforcement learning." 2019 international conference on robotics and automation (ICRA). IEEE, 2019.