Here's the first in a series of videos demonstrating the functionality of the platform. We go from recording data with the vehicle using engine_ai, uploading the logs to a server and training with trainer_ai, to deploying the result back to the vehicle for a live drive test.
Below we present a two-part series on the iterative process we followed to train a network to follow a pedestrian using the platform.
In the first session we present the base network implemented for the task and review the resulting drive tests.
In this second session we iterate on the base network to produce a robust model and conclude with a fun drive test.
Visit our GitHUB !