Select the Right Combination of Sensors to Match Application Needs and Use Multi‑Sensor Fusion to Improve Reliability.
Integrated Perception Systems Can Overcome the Deficiencies of Individual Sensors.
Nationwide, United States
Perception Technologies for Autonomous Transportation: A Comparative Analysis of LiDAR, Radar, Camera, and Sonar
Summary Information
Perception technologies, such as Light Detection and Ranging (LiDAR), Radio Detection and Ranging (Radar), Camera, and Sound Navigation and Ranging (Sonar) are core sensing technologies in modern intelligent transportation systems and automated vehicles. Multi-sensor fusion in different combinations has the potential to leverage each technology’s strengths and be applied for specific situations. This study applied a cross-modal analysis to directly link sensor performance metrics to specific transportation application requirements. It provided a comparative evaluation and examined the benefits and challenges of multi-sensor fusion strategies.
Some key considerations for sensor fusion included the following:
- Select the right combination of sensors to match application needs. No single sensing technology can provide geometric accuracy, semantic understanding, weather robustness, and low-cost near-field detection at the same time. Reliable systems should combine sensors whose strengths offset one another while balancing performance against cost constraints.
- Use multi‑sensor fusion to improve reliability. Combining modalities, such as LiDAR’s geometric accuracy with a camera’s semantic understanding or the weather resistance of radar can yield a perception system that is more robust than the sum of its parts.
- Match the sensor selection to the task. Sensor selection should be driven by the operational need. Radar is well suited to safety-critical range and speed functions, cameras to scene interpretation, LiDAR to precise 3D perception, and sonar to short-range low-speed tasks
