Context: solving the indoor navigation problem.
Decision: a neural network model of trajectory prediction is constructed.
Results:
The project was carried out, the solution was patented by the customer:
- Solution error: less than 5% of the distance travelled;
- The neural network model is patented;
- Two scientific articles were written.
To build the model, we used:
- Wearable device data;
- Information about the location of the wearable device;
- Ground Truth trajectories of human movement.
Simulation result:
- 2D trajectory of human movement.
Customer: Electronics manufacturer.
Technology stack: Python, PyTorch.