Ankur Srivastava

Topic: Study and implement the concept to support Recurrent Neural Networks on embedded hardware for Autonomous Vehicles. Tasks:- 1. Literature Survey on the adaptation of RNNs in ADAS to solve problems like Semantic Segmentation, Driver Intent/ Behavior prediction, etc. 2. Study the implementation of LSTMs in the existing hardware parallel co-processor. 3. Study the potential changes in the current hardware accelerators to create new IP dedicated to support RNN efficiently.

 Development and support for implementing Deep Learning Algorithms from high level frameworks (Caffe, Keras, TensorFlow) into embedded Renesas Devices.  Evaluation of different Object Detection Algorithms.  Designing and optimizing the existing object detection architectures for optimum use of hardware accelerators.

• Plan and organize the maintenance team. • Troubleshooting of Hydraulic and Pneumatic equipment and, to direct the respective shift or general team of a particular line to its respective machinery, equipment, electrical and mechanical systems. • Preparation of reports and statistics related to the critical machines or bottle neck.

Deep learning, Robotics and Embedded Systems