Visit Variscite’s Software Wiki for a full guide on how to build Yocto from source code. Yocto Dunfell 3.1 (Based on FSL Community BSP). This document uses Variscite's Yocto Dunfell BSP as an example: Yocto Project GPU: Vivante™ GC7000Lite 3D and GC520L 2Dįor more information, visit Variscite’s DART-MX8M-MINI Software Wiki and Product Page This document uses the Variscite DART-MX8M-MINI module as an example: DART-MX8M-MINI Creating a new meta layer and modify it to be a meta BSP layer with a machine configuration file: 1.2. Creating a new meta layer on top of Variscite Yocto BSP:ģ. Preparing the environment to build Yocto:Ģ. The document is divided into three parts:ġ. Then, you will modify it to be a BSP layer by adding a machine configuration file. You will learn how to create a meta layer that compiles and installs a 'Hello, World!' application. The same can be said to the training steps, and it is very dependent of your reward signal, on the same past example with a bad reward, it learned nothing after 40m steps, while it was doing ok after 500k steps with a good reward.This article demonstrates how to create a Yocto BSP layer for a custom carrier board designed for a Variscite System on Module / Computer on Module. For simple tasks like the agent collecting boxes using visual obsevation, I used two layers with 512 units each and it worked nice, but the best thing you can do is test various layouts. The confusing part is that it is not just the number of inputs (vectors or pixels) but the underlying information you want to learn. Let's assume it is a view from top of your map, most parts of it will be constant or rarely change (even more if it is downsampled to 84x84) and for learning we need variation in the observations.įor the model size, take this with a good pinch of salt, I believe it must be somewhat proportional to the state space and the desired behavior. I wouldn't call the convolution and downsampling process a compression because it is not exactly reversible, but more like a (learned) mapping from the image to features.Ībout your example I cant say for sure, but I dont think a mini-map is good. But use your tries to learn how to develop better agents, environments and specially rewards. Dont take general conclusions from your first (or various) try. In the end, the truth is that it all depends of your case and implementation, a single modification can take you from zero-to-hero and vice-versa. The downside is that training and inferencing with visual observation takes a lot of time. I had one case where the visual observation was much better than vector observation (about 2x the performance). With bad feedback on its actions it will learn nothing be it vector, visual or any kind of observation. The rewards are the signals that teaches it which actions connects with which observations, it is a lot more important than the kind of observation as long as it contains the needed information. Three important things about an agent are: What it observes, what actions it can take, and the most important how you reward the good and bad actions. Using vector observation versus visual observations is very debatable, it may be easier to the neural network build its own undertanding from the world using visual observation than learning human concepts and its encoding with vector observations. It is pretty easy to blow up all your GPU RAM depending of the buffer_size and image size you are using.Ĭhosing Gray vs RGB is very dependent of your use case, color information can be very valuable for example in a autonomous car in a road with yellow strips and green vegetation in the margins, dont underestimate it. ![]() Also remember that you collect #buffer_size images before training and it need to fit in your RAM or GPU RAM. About the visual observation itself, stick with the 84x84 size making sure that at this size the desired information can be "seen".
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |