Autonomous Vehicles
Also known as: AV, Autonomous Vehicle
Definition
Autonomous Vehicles are physical vehicles equipped with onboard sensing, computer vision, and decision-making systems that allow them to navigate, operate, or perform tasks with reduced or no direct human control, spanning applications including autonomous haul trucks, uncrewed surface vessels, and unmanned aircraft.
Key Points
- Autonomous vehicles are a robotics application — physical actuation combined with onboard perception and decision-making — and rely on computer vision and anomaly detection running via edge inference to operate independently of continuous network connectivity.
- The connectivity requirement for an autonomous vehicle is inversely related to its onboard autonomy level: a highly autonomous vehicle needs only intermittent connectivity for oversight and updates, while a lower-autonomy, remotely-supervised vehicle needs a much more persistent, low-latency link.
- Fleet-level coordination of autonomous vehicles — collision avoidance between multiple autonomous haul trucks, traffic management for autonomous vessels in a port — typically requires a network layer distinct from each vehicle's own onboard perception system.
- Safety-critical autonomous vehicle functions are engineered to fail safe if connectivity is lost, since the vehicle cannot be permitted to depend on a live network link for basic collision avoidance.
- Autonomous vehicle deployments generate substantial telemetry and sensor log data, which is typically transmitted opportunistically (at a depot, port, or gate) rather than continuously, given the bandwidth cost of raw sensor data relative to onboard processing capacity.
Concept
An autonomous vehicle's operational independence rests on its ability to perceive its environment (via cameras, radar, lidar, and other sensors) and make navigation or task decisions locally, using onboard edge inference rather than a continuous connection to a remote system. This is what allows autonomous vehicles to continue operating through connectivity gaps that would stall a purely remote-controlled vehicle. However, autonomy does not eliminate the network dependency — it relocates it. Fleet oversight, exception escalation, remote intervention when the vehicle encounters a scenario outside its operational envelope, and periodic software or model updates all still require a network path back to a remote operations centre or fleet management system.
Explainer
For ConnectedEarth's audience, the connectivity story around autonomous vehicles is about the shift from continuous remote control to intermittent supervisory connectivity. A mining company deploying autonomous haul trucks does not need a persistent, low-latency link to every truck — each truck's onboard computer vision and anomaly detection systems handle real-time navigation and obstacle avoidance independently. What the mine does need is a reliable link to a remote operations centre for fleet coordination, safety overrides, and exception handling when a truck encounters an unplanned condition. The same pattern holds for autonomous vessels navigating between ports and unmanned aircraft conducting surveying missions: connectivity requirements shift from continuous control-loop bandwidth toward periodic, supervisory, and exception-driven data exchange, changing the network architecture required to support the fleet from what a remotely piloted equivalent would need.