6 ms·
How do you train a model to drive with LiDAR when the human drivers who generate the training data don’t use LiDAR?
by Tycho 9mo ago
How do you train a model to drive with LiDAR when the human drivers who generate the training data don’t use LiDAR?
- knallfrosch 9mo agoScan with LiDAR while manually driving.
- jayd16 9mo agoHell, you could even use slower offline 3d reconstruction of vision data for training, and still ultimately rely on runtime LiDAR.
- Tycho 9mo agoBut the driver isn’t reacting to any of the LiDAR readings, only what they can see, so what is the point?
- aniviacat 9mo agoYou label the recordings that involve a crash, and use the lidar to avoid doing the same.
- wongarsu 9mo agoMy impression was that the state of the are was still to generate high-level data from your inputs, then react with a mixture of ML and algorithmic rules to those inputs. For example you'd use a mix of LIDAR and vision to detect that there's a pedestrian, use past frames and ML to predict the pedestrian's next position, then algorithmically check whether your vehicle's path is likely to intersect with the pedestrian's path and take appropriate action if that's the case Under that model, LIDAR training data is easy to generate. Create situations in a lab or take recordings from real drives, label them with the high-level information contained in them and train your models to extract it. Making use of that information is the next step but doesn't fundamentally change with your sensor choice, apart from the amount of information available at different speeds, distances and driving conditions
- 8note 9mo agosimulate is one way. put the car in a video game and raytrace what the lidar would see