![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F344232%2F2848960c9f3fd3610ac2d4d65f4e05ce%2Ffrank.png?generation=1600590898565350&alt=media\")
There are two approaches very popular amongst self driving car engineers to build such vehicles.
\nThe first one is the Robotics Approach-Robotics approach makes use of various sensors like lidar, sonar, GPS, radar, odometer, etc to enable vehicles to make a blueprint of the environment in which the car is driving and enables it to make predictions on how to drive the vehicles.This approach uses hard coding methods.
\nThe second one is Computer Vision and Deep Learning Approach-This approach basically uses several cameras feeds as inputs and the measurement of steering angles, speed, throttle, brakes, etc and the model learns to clone the human behaviour on how to drive the vehicle.
There are two approaches very popular amongst self driving car engineers to build such vehicles.
\nThe first one is the Robotics Approach-Robotics approach makes use of various sensors like lidar, sonar, GPS, radar, odometer, etc to enable vehicles to make a blueprint of the environment in which the car is driving and enables it to make predictions on how to drive the vehicles.This approach uses hard coding methods.
\nThe second one is Computer Vision and Deep Learning Approach-This approach basically uses several cameras feeds as inputs and the measurement of steering angles, speed, throttle, brakes, etc and the model learns to clone the human behaviour on how to drive the vehicle.
Thanks for sharing this information. Have a great day ahead!
", "rawMarkdown": "Thanks for sharing this information. Have a great day ahead!", "votes": null }, { "id": "1015334", "postDate": "09/18/2020 05:58:44", "content": "", "rawMarkdown": "rishikeshkanabar 👍👍", "votes": null }, { "id": "1015862", "postDate": "09/18/2020 13:42:57", "content": "@penchalaiah123 Dont you think second one is also robotic approach? I would choose rule based approach and learning based approach as apt description of two types.
", "rawMarkdown": "penchalaiah123 Dont you think second one is also robotic approach? I would choose rule based approach and learning based approach as apt description of two types.", "votes": null }, { "id": "1016373", "postDate": "09/18/2020 22:27:14", "content": "Tesla uses a nature camera and has the best algorithm for AV now. Some company use Lidar but cannot archive the same
", "rawMarkdown": "Tesla uses a nature camera and has the best algorithm for AV now. Some company use Lidar but cannot archive the same", "votes": null }, { "id": "1018778", "postDate": "09/20/2020 01:01:18", "content": "I am not an expert in this subject but I would think the combination of two approaches (learning based + rule based for override in some specific cases) would make sense to me in my humble opinion.
", "rawMarkdown": "I am not an expert in this subject but I would think the combination of two approaches (learning based + rule based for override in some specific cases) would make sense to me in my humble opinion.", "votes": null }, { "id": "1019166", "postDate": "09/20/2020 08:35:45", "content": "Ya I think thats what this New image from Lyft medium article talks about. You prolly are right :) Amount of learning based solutions keeps decreasing downstream. Learning based to generalize task and Rule based to add robustness and system constraints
Which one will make the shot and be dominating in future? I assume the second one but it could be combined version of both, does that exist already?
", "rawMarkdown": "Which one will make the shot and be dominating in future? I assume the second one but it could be combined version of both, does that exist already?", "votes": null } ], "comments": [ { "id": 1015332, "author_name": "rishikeshkanabar", "author_url": "", "post_date": "09/18/2020 05:57:17", "content": "Thanks for sharing this information. Have a great day ahead!
", "votes": null, "replies": [] }, { "id": 1015334, "author_name": "penchalaiah123", "author_url": "", "post_date": "09/18/2020 05:58:44", "content": "", "votes": null, "replies": [] }, { "id": 1015862, "author_name": "deepakrajpurushothaman", "author_url": "", "post_date": "09/18/2020 13:42:57", "content": "@penchalaiah123 Dont you think second one is also robotic approach? I would choose rule based approach and learning based approach as apt description of two types.
", "votes": null, "replies": [ { "id": 1018778, "author_name": "pukkinming", "author_url": "", "post_date": "09/20/2020 01:01:18", "content": "I am not an expert in this subject but I would think the combination of two approaches (learning based + rule based for override in some specific cases) would make sense to me in my humble opinion.
", "votes": null, "replies": [] }, { "id": 1019166, "author_name": "deepakrajpurushothaman", "author_url": "", "post_date": "09/20/2020 08:35:45", "content": "Ya I think thats what this New image from Lyft medium article talks about. You prolly are right :) Amount of learning based solutions keeps decreasing downstream. Learning based to generalize task and Rule based to add robustness and system constraints
Tesla uses a nature camera and has the best algorithm for AV now. Some company use Lidar but cannot archive the same
", "votes": null, "replies": [] }, { "id": 1019833, "author_name": "aliabdin1", "author_url": "", "post_date": "09/20/2020 17:31:31", "content": "Which one will make the shot and be dominating in future? I assume the second one but it could be combined version of both, does that exist already?
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1015291": "There are two approaches very popular amongst self driving car engineers to build such vehicles.\nThe first one is the Robotics Approach-Robotics approach makes use of various sensors like lidar, sonar, GPS, radar, odometer, etc to enable vehicles to make a blueprint of the environment in which the car is driving and enables it to make predictions on how to drive the vehicles.This approach uses hard coding methods.\nThe second one is Computer Vision and Deep Learning Approach-This approach basically uses several cameras feeds as inputs and the measurement of steering angles, speed, throttle, brakes, etc and the model learns to clone the human behaviour on how to drive the vehicle.", "1015332": "Thanks for sharing this information. Have a great day ahead!", "1015334": "rishikeshkanabar 👍👍", "1015862": "penchalaiah123 Dont you think second one is also robotic approach? I would choose rule based approach and learning based approach as apt description of two types.", "1016373": "Tesla uses a nature camera and has the best algorithm for AV now. Some company use Lidar but cannot archive the same", "1018778": "I am not an expert in this subject but I would think the combination of two approaches (learning based + rule based for override in some specific cases) would make sense to me in my humble opinion.", "1019166": "Ya I think thats what this New image from Lyft medium article talks about. You prolly are right :) Amount of learning based solutions keeps decreasing downstream. Learning based to generalize task and Rule based to add robustness and system constraints ", "1019833": "Which one will make the shot and be dominating in future? I assume the second one but it could be combined version of both, does that exist already?" }, "source": "meta" }