The use of machine learning and artificial intelligence in autonomous (self-driving) cars for having smooth traffic, reduced fuel consumption and improved air quality may sound like a science fiction, but 2 test projects conducted by the researchers of the Berkeley Laboratory are showing positive results. Applying machine learning for transportation is a new application that is beneficial for both humans and the environment.

Along with UC Berkeley, Berkeley Lab scientists and researchers are using deep reinforcement learning, which is a computational tool used for training controllers to make transportation more viable and environment friendly. The first project uses deep reinforcement learning which will program autonomous vehicles to drive and simultaneously improve traffic flow, which reduces energy consumption.

The second project uses deep learning algorithms to analyze satellite images combined with traffic information from cell phones. The algorithm is tasked with using this data and an environmental sensor’s data to suggest ways to improve air quality.  

CIRCLES (Congestion Impact Reduction via CAV in the loop Lagrangian Energy Smoothing) is the traffic smoothing project led by Berkeley Lab researchers. CIRCLES is based on a software framework called Flow. This software framework allows researchers to discover and benchmark methods to optimize traffic flow. Flow can simulate the driving patterns of thousands of vehicles.

Deep reinforcement softwares are used to teach computers new concepts. It is also used to train computers to play chess or teach a robot to run over an obstacle. These program cars to check activities of the neighboring, so that it will try different types of actions, which include accelerating, decelerating or changing the lane.

By using all these sensors and softwares, algorithms are able to identify a speed which consumes the least amount of fuel, emit lesser greenhouse gases, they are also able to identify the best path for a car to move ahead in consonance with other cars on the road.

Most of the accidents which occur on the road, happen due to a human error. When a car is totally controlled by a software, the possibility of an error is less. These softwares allow a car to be entirely controlled by Artificial Intelligence, thus reducing the chances of an accident.

This ensures that we have

l Less fuel consumption

l Less air pollution

l Less noise pollution

l Less human intervention and accidents

The idea of using computers to drive cars, with little or no human intervention looks fantastic and incredible at this point of time. However, it’s important to note that there was a time when people considered cars a nuisance, but we have still come a long way since then. The same is applicable for AI and Machine Learning. It may sound utopian and outlandish, but it’s an idea whose time has come.

Machine Learning and Artificial Intelligence is here to stay. The real challenge is identifying every industry and sector that can be targeted and disrupted using

Machine learning. It is a technology that is here to stay. BSE Institute’s short term course on Machine Learning is a great option for students and professionals looking to start their careers in this industry. The course explains the technology and its practical applicability, which enables participants to assess and apply this technology in respective domains.