Chapter
Leaked Information in Human Robot Collaboration
1:19:12 - 1:26:48 (07:36)
The robot can collect information from external torques and can even plan its motions by disambiguating through E-stop signals to improve its ability to determine what the user wants.
Clips
By adding external torque to a robot's existing torque, a relatively optimal overall action can be produced, and the intervening of a safety driver in autonomous vehicles indicates a disagreement with the notion of optimality.
1:19:12 - 1:21:01 (01:48)
Summary
By adding external torque to a robot's existing torque, a relatively optimal overall action can be produced, and the intervening of a safety driver in autonomous vehicles indicates a disagreement with the notion of optimality.
ChapterLeaked Information in Human Robot Collaboration
Episode#81 – Anca Dragan: Human-Robot Interaction and Reward Engineering
PodcastLex Fridman Podcast
The way a robot moves when an emergency stop is pressed can provide information on how it should move in the future.
1:21:01 - 1:24:16 (03:15)
Summary
The way a robot moves when an emergency stop is pressed can provide information on how it should move in the future. Planning robots' motion is necessary to avoid issues while trying to do tasks.
ChapterLeaked Information in Human Robot Collaboration
Episode#81 – Anca Dragan: Human-Robot Interaction and Reward Engineering
PodcastLex Fridman Podcast
The reward function in robotics tells a robot what behavior is good behavior and what it should be optimizing for.
1:24:16 - 1:26:48 (02:32)
Summary
The reward function in robotics tells a robot what behavior is good behavior and what it should be optimizing for. The robot can understand that even though it may not see a person's action, it can still figure out what is desired from the reward function.