Position and Velocity Control with LSM6DSV16x

[FKara.3]

Hello,

I am looking to use the LSM6DSV16x sensor for position and velocity control. What is the most effective way to achieve the best results? How can I utilize FSM, MLC, and sensor fusion to optimize performance?

To clarify, the purposes are:

- Trajectory following with mobile robot

- Velocity control in both straight and tilted ways. 

[RhSilicon]

This video below seems to be related to the use of this sensor, see if it helps:

Smart motion sensors are enabling distributed machine learning to significantly reduce bandwidth for more responsive, energy-conscious edge computing solutions. ST's latest ultra-low-power 6-axis inertial sensor (LSM6DSV16X) comes with a Machine Learning Core (MLC) and Finite State Machine (FSM) to enable motion pattern recognition or vibration detection. These smart inertial sensors have the capability to execute decision trees in the built-in MLC of the sensor which is ideal for always-on applications for wearables or wireless sensor nodes with a current consumption of only a few microamps.

To help companies stay ahead of the design curve and get their products to market quicker, powerful neural network conversion tools like our STM32Cube.AI make it easy to import and convert pre-trained Machine Learning or Neural Network models into optimized C code for STM32. In addition, middleware is available for easy integration with popular mobile platforms, such as Android, which are commonly used to build smart devices for consumer, industrial, and automotive applications.

https://www.youtube.com/watch?v=uBDYP6hp8PY

[niccolò · Accepted solution]

Hi @FKara.3 ,

this is a very interesting project, let me try to give you all the info you need:

if you need to detect and follow the motion of a robot in the space, the Sensor Fusion algorithm can help you in detecting the orientation of the robot, with a very high precision.
you can find all the info on it in this application note (section 6.5).

then when speaking about velocity, the situation is more complicated: in theory, having the acceleration, it is possible to integrate it to get the velocity, but in reality, there are many factors to keep into consideration, that makes it impossible to have a great accuracy.
the noise (both the environmental one and the sensor one), is a big problem, but with the right algorithm you can find a good compromise.

If this answers your question, please, mark this as "best answer", by clicking on the "accept as solution" to help the other users of the community

Niccolò