Tesla Autopilot Head's Support for Neural Networks Over LiDAR

author: FutureCar Staff    

Tesla Inc's humanoid bot Optimus, which relies on neural networks and cameras, is being praised by the company's executives. Ashok Elluswamy, Tesla's Director of Autopilot software, stated that this approach is the "solution" for robotics. He shared a video demonstrating Optimus' adaptability and precision, showcasing its ability to transition from sorting objects to performing yoga poses.

Elluswamy believes that neural networks and cameras are the future of robotics, and he predicts that it will become obvious in the coming years. He dismisses technologies like LiDAR, RADAR, ultrasonics, and HD maps as unnecessary. However, the bots also utilize additional sensing capabilities such as audio, proprioception, and temperature.

CEO Elon Musk also joined the discussion, mentioning upcoming improvements in Optimus' actuators, sensors, and mechanics. He even humorously suggested that the bot might "do the ballet one day." Musk has been a vocal critic of LiDAR, favoring cameras and neural networks instead. He argues that roads are designed for biological neural networks and eyes, making them more compatible with digital neural networks and cameras.

Tesla introduced Optimus two years ago and has created about 10 units. Musk announced plans for the first Optimus bot with Tesla-designed actuators to be ready by November. Tesla designs its own actuators as no supplier meets their requirements. Musk is confident that these bots will be performing useful tasks in Tesla factories by next year.

FutureCar Staff