Human-Like Behavior in Autonomous Systems
summary
Claims a concrete architecture that injects predictive digital twin models into vehicle-to-vehicle interaction, computes a blended trust score (0–100%) from direct simulation compliance plus reputation, and drives a baseline safety module via trust-aware decision-tree nodes. Implementation uses SUMO traffic simulation and student-built frameworks; simulation validation shows fewer incidents and roughly 30% throughput improvement in small scenarios. Components: digital-twin manager, trust manager, aggregator, adaptive safety module; validated in simulation, industry trials planned.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:52Problems and benefits of human drivers
- 5:01Autonomous ecosystems challenges
- 6:06Trust and safety in AVs
- 10:02Adaptive safety architecture
- 13:13Real-world validation
- 14:42Thesis progress and future
- 15:09Question and answer session
related talks
42mAutomated Driving - Taking the Human Factors Researcher out of the Loop
59mOn Human-robot Interactions via Multimodal Intent Modeling and Reachability-based Safety Assurance
25mDriver Models in SUMO and Elsewhere
27mPreparing for the Autonomous Vehicle Era
21mAutonomous driving and handling the handover
27mDriving Autonomous Vehicles Forward w/Real World Applications
57mReal-Time Decision Making in Hybrid Systems
21mFuture in Motion: Robotics, Self-Driving Cars, and Humanoids Shaping Tomorrow