isaacSimJetbotObjDet
Jupyter notebook implementation for a jetbot controller compacted in an object detection algorithm for Isaac Sim
isaac_sim_jetbot_objdet.ipynb - the script is meant to be used in connection with Isaac Sim simulator: It loads the scene with a cube, groundplane and the jetbot (objects provided in the script) It activates the camera from the viewport It reads the camera image and process through the deep neural network already trained for basic shape objects
How to: Run the simulation and go to Layer-> activate Live sync
./jupyter_notebook.sh isaac_sim_jetbot_objdet.ipynb
Open the link
isaac_sim_jetbot_objdet_usd.ipynb - same as previously with the only difference that the simulation is loaded directly from an .usd file already ready. #TODO - Implement dynamic control for jetbot acquired from the usd file.
jetbot_objdet_v2.ipynb - is meant to work for a real jetbot equipped with a camera: It activates the camera from Jetbot class (Camera) It reads the camera image and process through the deep neural network
simple_sphere.ipynb - notebook with procedures and steps to train the CNN SSD Mobilenet v1 Coco 2018_01_28 Number of classes: 1 Batch size: 24 Train steps: 50000 Training data: 800 Testing data: 200 Dependencies: Tensorflow 1.15, Numpy 1.19.5