Hazard Identification and Avoidance Simplified
RF-eye identification beacons on objects provides an orthogonal information system for Advanced Driving Systems which corroborate visual data. The anonymous beacons identify:
Road Users: ML algorithms have difficulty identifying pedestrians, especially children, bicycles, and other road users.
Transformation: Road objects or personal items embedded with RF beacons specify unambiguously identify objects along the road, protecting both the vehicle and other road users. Local, accurate data removes the requirement of connectivity for DNN inferences on camera data to identify potential hazards. RF-eye beacons train ML algorithms what road user objects look like in unusual conditions (fog, smoke, etc.,); over time, reducing “confusion” in autonomous vehicles.
Several incidences highlight the need for more accurate and timely object identification:
a woman was killed in an intersection after being knocked down then subsequently run over by an autonomous vehicle. Had the woman been wearing shoes identifying her as human, the vehicle would have identified the hazard and avoided killing her.
Cruise has had its permit for autonomous operation in California revoked, one reason being the difficulty in identifying children who are less predictable than adults.
