4 ms·
Elon Musk: “Anyone relying on Lidar is doomed.” Experts: Maybe not
- sabertoothed 7y agoClickbaiting headline. Hotz and Levandowski are also experts and they agree with Musk. More truthful headline: Some experts say, but then again other experts say.
- Analemma_ 7y agoHotz isn’t an expert, he’s a loudmouth tinkerer who never produced anything working before NHTSA shut him down. I’ve never heard Levandowski comment on the long-term necessity of lidar, but actions speak louder than words: Otto/Uber definitely used lidar (actually, the entire lawsuit with Waymo was over their lidar tech. Why would Levandowski have risked everything stealing it if he didn’t think it was necessary?) Elon really is alone in claiming that lidar isn’t needed for self-driving.
- sabertoothed 7y agoYou seem very angry for some reason that I do not understand. Your emotional response is not backed up by facts. Hotz can certainly be considered an expert as founder and president of commai.ai. Whether he is a loudmouth or not is less relevant. The remark by Levandowski can be found here: https://www.youtube.com/watch?v=fNgEG5rCav4 https://www.youtube.com/watch?v=fNgEG5rCav4 Whether these guys are ethical etc. is not of interest for this question. > Elon really is alone in claiming that lidar isn’t needed for self-driving. No, precisely not.
- checktheorder 7y agoThe article didn't seem to go into regulatory issues at all. I'm not qualified to speak to the technical merits of LIDAR versus computer vision systems. But coming from a business planning background, and considering the fact that global rules surrounding LIDAR's use in consumer products can be charitably called "a patchwork", it seems to me that computer vision systems would be far easier to implement in a product designed for global export than LIDAR systems.
- matt-attack 7y agoI actually agree and have been saying this for years. Cars need to model as much as possible on the way humans drive. No one will expect an autonomous car to be able to do things that an alert and expert driver can achieve. The entire road system is built around the premise that it can be navigated using vision alone, with no a priori knowledge of its layout. You can drop me onto a roadway undergoing construction with temporary lanes that are diverted to the left side of the road, all in a suburb of Boston, a city I’ve never been to, at night in a rain storm. You and I can safely drive in this condition if alert and focused, while relying only on vision. My premise is that any system that requires anything else, e.g. map data, lane boundary data, lidar, is doomed to fail in these edge case scenarios.