Next Generation Motion Sensors for Hybrid  GNSS INS Solutions in high accuracy machine control applications

Next Generation Motion Sensors for Hybrid GNSS INS Solutions in high accuracy machine control applications


Robotics, machine control, industry automation require high accuracy positioning, as well as very robust PNT to support autonomous operations. Relative motion sensors (e.g. IMU) are key to complement GNSS in case of outage or for enhanced robustness and sustained high accuracy. In open sky, the state-of-the-art of GNSS and IMU provide good performance. However, cost-efficient IMUs still face shortcomings in filling the gaps of long GNSS outages (urban, large factory plants, canopy, etc.) or in supporting very high-accuracy positioning.
Intense R&D efforts (e.g. DARPA) are invested to prepare the next generation of inertial navigation systems (INS) with the goal of achieving chip-scale sized sensors that simultaneously are able to meet the performances of tactical grade IMUs. For multiple reasons, the outcome of those investments might not reach the industrial and commercial markets until in 10 years or more, and the cost of high performance IMUs will remain high.
Meanwhile, composite motion sensors, which combine multiple low-cost technologies, could provide a relevant path to satisfy the growing need for costefficient, high-performance INS in domains such as robotics in industry automation, machine control and high-precision agriculture automation.
Typically, camera and mmWave radar are able to provide relevant motion measurements (odometry as well as attitude). Such technologies are individually reaching a good TRL level, the challenge lies in their cost-efficient integration, tuned to the adequate use cases, which requires advanced sensor fusion, for which Artificial Intelligence seems to be the most suitable technique to apply.

The objective of the proposed activity is the development of the next generation of motion sensors based on massive (low-cost) sensor fusion and Artificial Intelligence/Machine Learning, to enhance calibration and sensor fusion. Design and performance will be tuned for integration with GNSS in use cases requiring robust, high-accuracy positioning (e.g. robots used in Precision Agriculture or in logistics on large factory plants outdoors).
The tasks to be performed will include:

  • development and proof of concept of a disruptive composite motion sensor, merging multiple low cost technologies such as mmW radar, camera (attitude/odometry) and low cost MEMS (gyro-, accelero- and baro-meter); 
  • tuning for the targeted uses cases (e.g. machine / robot control operating in challenging environment and seeking robust high accuracy) both in term of multi-sensor architectures and AI-training; 
  • assessment and demonstration of multi-sensor End-to-End performance, representative of the targeted use cases.

The main results of the activity will provide:

  • a breadboard of the developed new composite motion sensor;
  • results of test trials and benchmark with state-of-the-art GNSS/IMU solutions in various environments; 
  • a roadmap towards prototyping and a product, paving the way for industrial products, for instance through activities in NAVISP Element 2.