Issue link: https://resources.mouser.com/i/1442859
12 | The Future of Robotics Texas Instruments Reliability Sensor fusion also leads to increased reliability. By its definition, Sensor fusion means that sensing is not relying on a single point of failure. When two or more sensors are employed and their data fused, data will still be received if one of the sensors failed. This does not prohibit a more catastrophic failure from disabling the system, such as an electric overload, or severe physical trauma or damage. However, it does mitigate from the component and offload the potential for component failure while still requiring an overall system fail-safe. Sensor fusion can also provide more reliable information because of voting. This means if several sensors are measuring the same thing, that is, they are all providing data, the system might select and employ the data from the senses that correspond the closest, believing this data representative of that which most closely corresponds to reality. Estimating Another goal that sensor fusion helps to achieve is proving useful estimating when something might not be directly measured because it is in the context of being temporarily unmeasurable. Often in these applications, a combination of sensor fusing added to the physical world's knowledge is brought together by way of a Kalman filter. A Kalman filter (1960) is a mathematical algorithm invented by Rudolf Emil Kálmán (1930-2016). It is commonly utilized in navigation, including Global Positioning System (GPS) (Figure 6). The Kalman filter makes a prediction and then gets an update. It uses a weighted average of the predicted value and the measured value and works to adjust itself to achieve the smallest linear difference. Mathematically, this provides excellent estimation because the weighted averaging approach smooths the estimate and adjusts course in tiny increments. Coverage Finally, sensor fusion can also be employed to increase the coverage area. The concept is simple. If one was looking for a fallen object in the field, one could employ a flashlight in the dark. However, a better solution would be to use many flashlights or include floodlights. In either instance, more light is thrown into the context, so the object becomes easier to locate and find. More sensors fused will provide more coverage. Conclusion Autonomous robotic systems require a fusion of diverse sensor data in near real-time. Sensor fusion marks the future of autonomous robotics systems by giving them the ability to sense, perceive, plan, and act. Gaining better data insights by way of sensor fusion is enabling autonomous robotic systems to achieve their goals. So, that one day, these autonomous robotic systems will be able to better sense and respond to their surrounding work environment like us. ■ Figure 6: Sensor fusion and Kalman filters help estimate navigation position, providing better insights from data. Learn More PGA460/PGA460-Q1 Ultrasonic Processor & Driver Learn More TUSS4470 Direct Drive Ultrasonic Sensor IC With Logarithmic Amplifier