How do autonomous vehicles make decisions?

Instruction: Outline the basic process from data collection to action in the context of an autonomous vehicle navigating a road.

Context: This question delves into the candidate's understanding of the decision-making algorithms that underpin autonomous vehicle behavior, encompassing data interpretation, planning, and execution.

Updated

Example Answer

A typical pipeline senses the environment, estimates objects and vehicle state, predicts plausible behavior, plans a feasible trajectory and controls actuators. Predictions are uncertain, so planning needs constraints and monitored limits rather than a single presumed future. Runtime feedback adjusts control and state estimates; that is different from changing learned model weights. Logged failures can inform offline development, with versioned validation and release approval before a new driving policy is deployed.

Make it your own

Illustrative boundary: control feedback updates steering commands using the deployed controller; training feedback changes model parameters only in a separately reviewed development release.

Why this works

This separates ordinary closed-loop control from ungated policy learning. Follow-up: what should the planner do when two plausible pedestrian trajectories conflict?

Sources: NHTSA ADS safety guidance, NHTSA vehicle cybersecurity practices.

Related Questions