Agents for safety-critical applications

Robert Moss, a PhD student in SISL, discusses how POMDPs are applied to solve safety-critical problems.

0:00 Introduction
0:30 Agents
1:11 Aviation safety applications
1:47 Safe autonomous navigation applications
2:27 Financial and cyber safety applications
2:48 Safe geological sustainability applications
3:44 MDP: Markov decision process
4:45 POMDP: Partially observable Markov decision process
5:38 Objective: Maximize expected discounted future rewards
6:40 Aircraft collision avoidance systems
10:28 Safey of autonomous cargo aircraft
14:13 Airborne wildfire suppression
16:56 Autonomous vehicle risk assessment
17:43 Autonomous lunar rover exploration
19:10 Critical mineral exploration
20:34 Algorithms for long-horizon sequential planning (BetaZero)
24:52 Why does this all matter?
25:43 Geothermal energy production POMDP