Experimental Assessment of UWB and Vision-Based Car Cooperative Positioning System

The availability of global navigation satellite systems (GNSS) on consumer devices has caused a dramatic change in every-day life and human behaviour globally. Although GNSS generally performs well outdoors, unavailability, intentional unintentional threats, reliability issues still remain. This motivated the deployment other complementary sensors such way that enables reliable positioning, even GNSS-challenged environments. Besides sensor integration single platform to remedy lack GNSS, data sharing between platforms, as collaborative offers further performance improvements for positioning. An essential element this approach is internode measurements, which brings strength geometric network. There are many can support ranging LiDAR, camera, radar, RF technologies, including UWB, LoRA, 5G, etc. In paper, demonstrate potential positioning technique, we use ultra-wide band (UWB) transceivers vision compensate unavailability terrestrial vehicle urban scenario. particular, cooperative exploiting both vehicle-to-infrastructure (V2I) vehicle-to-vehicle (V2V) UWB measurements have been developed tested an experiment involving four cars. results show be effectively used determine distances vehicles (at sub-meter level), their relative positions, especially when or sufficient number V2V ranges available. presence NLOS observations one principal factors causing decrease performance, but modern machine learning tools shown effective partially eliminating observations. According obtained results, V2I achieve level accuracy 2D not Combining may lead similar Absolute group requires stable certain provided with data. Results meter-level achieved at least two network usually receive 2–3 vehicles. These working conditions typically ensure robustness solution against undefined rotations. led accuracy, improvement absolute reduction required one.