Tesla autopilot now enables the car to perceive the space around it


Tesla Autopilot is now enabling the car to perceive the space around it thanks to the development of its occupancy grids. Tesla’s Director of Autopilot Programs, Ashok Elluswamy, shared a detailed thread on Twitter about the recent workshop the Autopilot team held. He also shared the workshop on Twitter.

In the video and Twitter Thread, Ashok explained how Tesla developed occupancy grids to literally give the car a sense of its surroundings. Humans have the ability to understand things around them at any time. Is this car traveling on the road at a slow or fast speed? Do I, as a pedestrian, have enough time to cross the street before I get hit? What is this in the middle of the road? What is this fallen from heaven? I have to get out of the way.

These reactions to scenarios and quick decisions come naturally to humans. Tesla’s autopilot team is working on programming the vehicles to do the same and that should save lives. Imagine that the car is able to correctly detect its surroundings while the driver is not even paying attention. An example is sudden unintended acceleration (SUA). Ashok noted that autopilot prevents about 40 of these types of accidents per day.

The workshop took place in June at this year’s conference on Computer Vision and Pattern Recognition (CVPR.) in New Orleans. Ashok explained that the team developed occupancy grids that enable the car to predict the volumetric occupancy of everything around it.

Ashok explained that typical methods such as segmentation of the image space for free space or pixel depth have many problems. The solution to these issues is occupancy networks.

In other words, occupancy networks enable a vehicle to be aware of the space around it and to determine whether or not it can drive in that space. For example, if a UFO suddenly crashes in front of you while you are driving, you will react quickly and in the safest way possible. This is what the Autopilot team trains the software to do.

Ashok shared details of how occupancy networks use neural radiation fields (NeRFs). “The occupancy representation of these networks allows for differential rendering of the images (based on the action of neural radiation fields). However, unlike typical NeRFs, which are present in each scene, these occupancy networks are generalized across scenes.”

You can read Ashok book The full Twitter thread is here You can view its presentation here. It’s been just over a month since Tesla’s AI Day and I’m sure Tesla will be sharing more about the life-saving technology it’s working on as well as the Optimus Bot.

Posted by Dr. See it first recently a video about the new 10.69 update and share his thoughts on Occupancy Network.

Lee said in a Twitter message, “The beauty of occupancy networks is that the car doesn’t have to know what or what The things he sees, he just has to know they are be There in order to avoid them! ”

Note: Johnna is a contributor to Tesla and supports its mission.

Your feedback is important. If you have any comments, concerns or see a misspelling, you can email me [email protected]. You can also reach me on Twitter Tweet embed

Tesla autopilot now enables the car to perceive the space around it






Leave a Reply

%d bloggers like this: