Table 4 Average episode runtime (seconds) for tasks differentiated by complexity. A* is computed as average time/step.

From: Robot movement planning for obstacle avoidance using reinforcement learning

Agents

Simple tasks

Complex tasks without wall

Complex tasks with wall

DDPG-APF

1.051

1.425

1.460

Pure-DDPG

1.225

1.526

1.515

TD3-APF

1.543

1.833

1.859

A* Search

1.375

1.924

2.067

Back to article page