You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First of all, I would like to thank the authors for sharing :). When using Interfuser to conduct related experiments, I found a very special situation. When the defined task distance is too long (the task is customized by me, i.e., the distance between the start and destination is very far), Interfuser cannot complete the task, always It stops in the middle (between the start point and destination point). I also tried the route in route_town05_short.xml and the same problem happened. I don’t know if Interfuer training scenarios are limited, or if there are better trained models that can be supported. Thank you :).
The text was updated successfully, but these errors were encountered:
Hi,
Thank you for your interest in our project. A practical workaround is to design some rules manually in the interfuser_agent.py to let the agent move if its speed is very low for a long time. You also adjust the controller parameters in the controller.py . To solve this problem totally, you may need to rewrite the autopilot agent to collect better codes. Some related issues can refer to: #71
First of all, I would like to thank the authors for sharing :). When using Interfuser to conduct related experiments, I found a very special situation. When the defined task distance is too long (the task is customized by me, i.e., the distance between the start and destination is very far), Interfuser cannot complete the task, always It stops in the middle (between the start point and destination point). I also tried the route in route_town05_short.xml and the same problem happened. I don’t know if Interfuer training scenarios are limited, or if there are better trained models that can be supported. Thank you :).
The text was updated successfully, but these errors were encountered: