Artificial Intelligence has gained lot of traction and importance in the 21st century with use cases ranging from speech recognition, learning, planning, problem solving to search engines etc. Artificial Intelligence also has played a key role in the development of autonomous vehicles and robots ranging from perception, localization, decision to controls. Within the big AI umbrella there is machine learning which is all about using your computer to "learn" how to deal with problems without “programming". Deep learning is a branch of machine learning based on a set of algorithms that learn to represent the data directly from the input such as an image, text, Sound, etc. Within deep learning there are Convolutional Neural Networks and Recurrent Neural Networks (CNN/RNN). The study here used convolutional neural network approach to perform image/object recognition. Given that the objective of the autonomous or semi-autonomous vehicle is to promote safety and reduce number of accidents, it is very important that the perception system is extremely robust for the systems downstream such as decision, path planning and control of the autonomous vehicle. The technique of Heterogeneous Convolutional Neural Network (HCNN) presented here has also been patented by the author that shows further benefits in terms of computational efficiency improvement and memory requirements. The downstream benefits of which reduces cost, increases the speed of the throughput without sacrificing accuracy.
Ensuring the safety of vehicle occupants has always been the primary focus of automakers. To achieve this goal, they have invested in the development of active safety features, which are designed to prevent accidents from occurring in the first place. These innovations are driven by a desire to save lives and reduce the risk of injury or death on the road. The implementation of advanced driver assistance systems (ADAS) and automated driving functions requires a high level of complexity and coordination between various subsystems. To meet these challenges, the overall autonomous system is divided into three main areas namely 1) perception 2) decision and 3) path planning and controls. The robustness of perception system to be able to recognize relevant images and objects under all operating conditions is one of the fundamentals of controlling the vehicle longitudinally and laterally to reduce number of accidents save lives/injuries. In this paper, we present a summary of our research on the use of the convolutional neural network to design perception systems for autonomous vehicles.