Abdul Jabbar

How occupancy grids, Bayesian filters, and particle-based models quantify risk in occluded space and produce safe motion constraints for planning.

HD maps, semantic priors, and encoded traffic rules turn unknown space into constrained hypotheses—helping planners avoid false negatives and pick conservative maneuvers at occluded intersections.

Practical maneuvers an autonomous vehicle can execute to improve its viewpoint and reduce uncertainty at occluded intersections, with trade-offs for safety, comfort, and traffic flow.

How AVs infer hidden hazards from the motions of nearby road users—braking, steering, signaling—and how to integrate those cues into prediction and planning.

Practical guidance and measurable checks teams can use to confirm synthetic edge cases reflect real-world failure modes and improve sim‑to‑real confidence.

How to engineer and evaluate simulators so they reproduce the causal interactions—occlusions, agent intent, and rare maneuvers—that determine autonomous-vehicle behavior in the real world.

Practical guidance for encoding kinematic limits, attention/perception models, decision rules and traffic-law constraints so simulated pedestrians and drivers behave plausibly.

How to build, run, and evaluate simulator-focused benchmark suites that stress low-frequency, high-risk driving events so teams can measure and close transfer gaps.

A practical taxonomy of edge-case families to guide data generation, test selection, and validation for AV stacks.

Practical steps for combining simulator training with targeted on‑road validation, shadow‑mode collection, and risk‑minimized test plans so simulated improvements transfer to real traffic.