Robot Navigation in Crowd Based on Dual Social Attention Deep Reinforcement Learning

Table 1

Results in the invisible setting.

Methods	Success	Collison	Time (s)	Reward
ORCA	0.43	0.57	10.86	0.0483
CADRL	0.87	0.12	11.14	0.2511
LSTM-RL	0.97	0.02	11.92	0.2820
SARL	0.99	0.01	10.49	0.3351
DSARL (our)	1.00	0.00	10.39	0.3439