Autonomous Driving

Andrew, Stanford 

Topics: Neural Network, Alvin, Automotive, IT Industry, Autonomous Systems 

Transcript Excerpt

Self-driving cars are rapidly becoming the norm on our roads. While the technology still has a long way to come the IT industry is working on creating them sooner rather than later. In this video, it is demonstrated how to create a self-driving car using a neural network and autonomous systems.  The goal is to revolutionize the automotive industry as self-driving cars will change how we interact with vehicles. The speaker shows a picture that demonstrates how neural networks will work in autonomous vehicles. The bottom left shows a video feed of the road as the car is driving. This video feed is then turned into a series of steering wheel inputs that are measured in confidence levels. These confidence levels are then read by the neural network to decide on which direction to turn the steering wheel, for how long and how sharp. These confidence levels are displayed by blue and white bars on the top left of the screen. The speaker highlights where these bars are and what they do. While the machine has trouble at first these kinds of systems can easily adapt to driving on a road all by themselves. Even when road conditions change these networks should be able to adapt based on what was previously learned.   

After this introduction, the speaker shows a video about Alvin which is a neural network that has learned how to drive itself through a course. The video starts with a human driver teaching the machine how to drive. Afterward, Alvin takes over and has to learn how to get through the course all by themselves. Just like the introduction, the machine measures how to drive through mimicking and confidence levels that the team can easily track. At first, the machine is unable to drive as it turns at random intervals. After around two minutes the car begins to learn how to properly drive and beings to make proper steering inputs. The car is now successfully driving. However, there is a massive challenge ahead. While the car has been driving on a single-lane road this whole time it comes to an intersection that leads onto a two-lane road. While Alvin’s confidence levels are lowered at the intersection it is still able to cross the intersection with no issues. When the machine goes on a two-lane road it is able to adapt to these new conditions to drive successfully still. This program shows the potential future of autonomous vehicles.       

Smart Waste Management Solution Using IoT

Autonomous Gripper System