Saturday at 18:331 day I need a challenge !3d three.js engine done , super wall click done , custom tv remote almost done .These aren’t 100% complete just requiring parts / 3d printing / motivation- but clearly no technical challenges left .So what next ? A full articulate human robot ? - issue then is vision recognition - that’s a challenge beyond current capabilities!So what next ? Me , Claude ultimate combo - I need new vision , new high bar , next level accomplishment.
Saturday at 18:551 day Author Kind of thought of this before ….https://mujoco.readthedocs.io/en/stable/index.htmlIt might have to be ….
Saturday at 21:391 day Author Robot it is !Start with jetson nano orin and oak lite camera . So object recognition and learning first ! Must try not to scale project to early … 🤣
21 hours ago21 hr Author Decided no need for Jetson. Also sent m5 back for refund. So PARSEC will leverage the m3 ultra as its brain via wifi.The oak camera should be interesting; got an older edition off ebay for 60 quid.Will use my mac air for testing first with camera with compute done on m3.Later will use a pi or nuc. - as its task is just to stream camera images.OAK-D Lite is the ultimate camera for robotic vision that perceives the world like a human by combining a stereo depth camera and a high-resolution color camera with on-device Neural Network inferencing and Computer Vision capabilities.Key Features:4 TOPS of processing power13 MP central RGB camera available in auto-focus and fixed-focus variantsBMI270 IMU, 6-axis sensor combining accelerometer and gyroscopeUSB2 / USB3 for power delivery and communicationRVC2 architecture on-board features:4 TOPS of processing power (1.4 TOPS for AI - RVC2 NN Performance)Run any AI model, even custom architected/built ones (models need to be converted)Encoding: H.264, H.265, MJPEG - 4K/30FPS, 1080P/60FPSComputer vision: warp (undistortion), resize, crop via ImageManip node, edge detection, feature tracking. You can also run custom CV functionsStereo depth perception with filtering, post-processing, RGB-depth alignment, and high configurabilityObject tracking: 2D and 3D tracking with ObjectTracker nodeBase line: 75 mmIdeal depth range: 40cm - 8m. See technical specifications.Maximum effective depth range: 20mPower consumption: Up to 5W. See technical specificationsIndustrial grade aluminum housing with Front Gorilla Glass1/4" tripod mount and 75mm M4 Vesa mountDimensions & Weight
20 hours ago20 hr Author 1 hour ago, -rick- said:You know about ROS?https://www.ros.org/I didn't no. I then mentioned it to Claude why this had been missed as it's clearly of considerable use. It said we are at Stage 1; Stage 2 is ROS - (expletive deleted)ing liar I thought!Looking back in the long Claude conversation I do now see "ROS" mentioned ... but its like in a list of stages with no explanation or detail.So Claude was correct! as it frequently is of course. Edited 20 hours ago20 hr by Pocster Ai showed me up
Create an account or sign in to comment