Risk-bounded control barrier functions replace martingale-based safety control to keep autonomous vehicles on planned trajectories with minimal adjustment.
Scenario-specific scoring functions classify driving context and rank candidate trajectories to improve autonomous vehicle path selection.
Danger zones trigger low-delay camera mode so moving bodies receive faster hazard alerts without sending all images at high frame rates.
Multiple onboard control modes let a leg-wheel hybrid vehicle handle sign, follow, and touch commands for omnidirectional movement in diverse scenarios.
Neural networks predict lane centeredness and road angle from camera images to steer autonomous vehicles through sharp corners without costly sensors.
When onboard autonomy meets road anomalies, remote expert takeover and live rule distribution maintain safe navigation without driver intervention.
Detects ride smoothness interruptions, traces the responsible control component, and adjusts jerk, acceleration, or planning parameters.
By spacing vehicles to allow 180-degree pivot turns, the platoon can leave dead ends without retreating, preserving comfort and efficiency.
Front wheel speeds are used to estimate rear wheel speed during cornering, improving brake and ABS control when rear sensing fails.
External sensors and edge planning improve autonomous vehicle reaction in bad weather and sudden events by sending timely route and control data.
By adjusting haul vehicle speed from predicted arrival gaps, this case cuts fuel use and waiting time at worksite loading points.
Camera-based obstacle detection adjusts corner radius and speed so mobile robots can turn aisle corners safely without losing travel efficiency.
Blending wheel speed with inertial sensing based on slip detection improves vehicle velocity estimation and limits integration drift.
Dynamic override thresholds and preselected off-path stop points help remote automated vehicles avoid intersection conflicts during risk mitigation.