AI & ML
D'BEAST: Logged Bad Experiences as Explanation for Autonomous Driving Agents
arXiv cs.RO0 views
arXiv:2611.20019v1 Announce Type: cross Abstract: Reinforcement learning agents learn from mistakes, yet their decisions can be difficult to trace back to those experiences. We investigate whether a tabular agent's decisions can be explained by its bad experiences, defined as the training transitions associated with safety or task events. A fixed training log is replayed from the same initial Q-table under three conditions. FULL retains all updates, SKIP omits bad-experience updates, and NEUTRALIZE retains those updates without their event penalties. We illustrate the method on six classic Markov decision processes and evaluate it in closed-loop driving in CARLA. On the full driving task, FULL reached the goal from 33 of 40 test starts, whereas SKIP and NEUTRALIZE reached it from none, indicating that the bad experiences carry decision-relevant information. At the local level, for 47 decisions shortly before a failure of SKIP or NEUTRALIZE, removing or neutralizing a small group of bad experiences, with a median size of one, caused the trained agent to select the failing policy's action in 46 cases. For this agent, the results indicate that its decisions can be explained by its bad experiences.
Source:https://arxiv.org/abs/2611.20019PreviousReactBench-AD: Do Driving VLAs and World-Action Models React to Imminent Hazards in Their Path?NextA Geometric Observer for Visual-Inertial Odometry using Optical-Flow Constraints
Related posts
OpenAI's Coding Agents Now Finish 46% of 2–4 Hour Research Tasks Alone — But the Frontier Is Jagged
0dev.to
Previous-Token Prediction Based LLM Inversion for Prompt Reconstruction
0Hacker News
FailTrace – Reproduce a failed GitHub Actions run locally from its URL
0Hacker News
AI Will Not Make You Rich (2025)
0Hacker News
Wabi as a Health and Habits Companion: 5 Trackers You Can Build Inside a Chat
0dev.to
Building Smarter Automotive Factories: How AI and Industrial IoT Work Together
0dev.to