Hi developers, I am new to this site and group. I have been developing a project for the last year or two based on my interest motorsports and AI. However, the software did not exist to process different data types. Then when GenAI came out, then RAG, and then Agentac AI project became feasible. Previously I had exposure to Jetson and Raspberry Pi systems and their increased computing power. My project is developing an edge AI system that resides in Motorsports race car collects sensor data, positional data, camera frames, and vehicle attitude dynamics (roll, pitch, yaw) for model training data. This data plus engine performance dyno results, track diagram, racing technical information will be used build a RAG/Agentic Model. This model will run as an embedded edge application on the fastest Jetson computer located in the dash of a 1996 Mustang. Power for the Jetsom will come from the car’s battery. The system will identify patterns of safety and performance information lap after lap. This information is input to the model, approximately every .1 second, and predicts whether potential safety or performance issues exist before they happen and the driver has time for corrective action. The system will alert the driver with correct action and the problem, The system is being designed to maximize parallel tasks, and similar to a process control computer with interrupt system, there are many offline motorsports sensor analysis system that are very costly. None can provide an AI racing engineer in the car during the race proving valuable information to the driver that they cannot obtain ed in any other way and at a very competitive cost. I have built a 1996 Mustang to use as a test lab that will be competitive in SCCA higher classes. The car has been racing on local tracks for the last two years. There is not much of the original Mustang left. Every component in the suspension adjustable along with engine tuning. The car has lost eight and is around 3100 lbs. with 480 HP at 7,500 ROM. The Jetson or multiple Jetsons will require the power for sub-second responses. I have developed a 20-page going into the what’s, how’s and why’s. This system will demonstrate the real advantage of AI is not replacing humans but enabling humans to achieve higher performance levels previously thought not possible. I welcome any feedback positive and negative. I believe we are building a game changer system based on Jetson and RAG/Agentic.
Hi,
Here is some information for your reference:
-
You can find our Jetson generative AI tutorial below:
Tutorials - Jetson AI Lab | Jetson AI Lab -
We have another product for automotive. Please also check the information below:
Autonomous AI Solutions | NVIDIA Developer
Thanks.