A key issue in the advancement of sensor fusion technology today is that the autonomous platform – car, robot, drone – must accurately assess the world around it and move accordingly, a complex endeavor. Despite advances made in navigation and guidance systems for unmanned and autonomous technologies, significant technical hurdles remain.
Gyroscopes, accelerometers, magnetometers, pressure sensors, optical systems, and other sensing devices acting individually have limits. Drawbacks associated with individual sensors include:
- Sensor deprivation – The breakdown of a sensor due to loss of signal, such as GNSS
- Limited spatial coverage – Individual sensors only cover a specific region or function
- Limited temporal coverage – Some sensors require a specific set-up time to perform and transmit measurements
By fusing data from multiple sensors – optical/camera, LiDAR, GPS, and inertial measurement units, such as those developed by KVH – system designers can use different sensing technologies to overcome the weaknesses of others and, in many cases, exponentially improve overall system performance. Collectively, position, direction, motion, velocity, and other measurements are more useful than when considered separately. Advantages of sensor fusion include:
- System robustness and reliability – Multiple sensors provide an inherent redundancy
- Extended spatial and temporal coverage – One or more sensors provide measurements even when others cannot
- Increased data confidence – Information from one sensor can be confirmed by another sensor covering the same domain
- Improved resolution – Combining information from multiple independent sensors creates a more accurate, higher resolution “picture”
While challenges remain, sensor fusion is increasingly considered key to the successful development of fully autonomous platforms, whether used on land, in the air, or at sea.
Discover the challenges and solutions to making unmanned and autonomous platforms as safe and effective as possible. Download the free ebook, “Unmanned Technology Challenges: Teaching Machines to Operate Like Humans” today.