<p>Shiv Nadar University is a comprehensive, multidisciplinary, research-focused, and student-centric University offering a full range of academic programs at the undergraduate, postgraduate and doctoral level. With state-of-the-art infrastructure, the University comprises of academic wings, sophisticated labs, international standard sports facilities, amphitheaters, auditorium, conference rooms and smart classrooms. The University&rsquo;s goal is to become internationally recognized for the quality of its research and creative endeavors and their applicability to improving quality of life, generating new insights and expanding the boundaries of human knowledge creativity. Committed to excellence in teaching, research and service, the University aims to serve the higher education needs of India and the world beyond.</p>

Shiv Nadar University

Reinforcement learning, a sub-field of Machine Learning has been garnering lot of research attention lately. It helps create intelligent agents that can incrementally learn optimal strategies for challenging environments by interacting with it. Such agents are best suited for solving problems like traffic congestion, which demand solutions that cater to dynamic changes in the traffic throughput. Intelligent transportation systems which use deep reinforcement learning can adapt to varying traffic demands and learn to maintain reduced congestion. In this paper, we propose a solution approach to use Double Deep Q Networks for traffic signal control of varied traffic flows in an isolated intersection. To improve the stability of our proposed method we have used target networks, delayed updates and experience replay mechanisms. We evaluate the performance of our method on different time-varying traffic flows and find that our method learns a robust and optimal strategy which reduces vehicle waiting time and queue length significantly. Our method achieved superior performance compared to traditional traffic signal control strategies. The method has been trained and evaluated through simulations of road networks created on Simulation of Urban Mobility (SUMO). © 2021 IEEE

World University root node

Electrical Engineering

(SoE) School of Engineering

Proceedings of the 8th International Conference on Signal Processing and Integrated Networks, SPIN 2021

Intelligent Traffic Control using Double Deep Q Networks for time-varying Traffic Flows

Nowadays robots have been inching towards living beings in terms of performance, agility, accuracy and performance. But there is one domain which is yet to come to limelight i.e. autonomous rescue robots. This involves a parallel use of robot design and control systems to get it executed. Basically, in simpler terms, there is a risk of life is involved in such scenarios and it's always hoped that a robot deployed for such operations be working with high precision and recall. So, in this paper we have described the design and control of an autonomous amphibian all terrain six-legged robot with adaptive gait for rescue operations. To fulfill this, we have used LiDAR and camera in parallel to assist bot in locomotion and detect the victim. Achieving all of this needs a significant amount of on-board processing power, but at the same time has to be compact enough. Keeping this in mind, NVIDIA Jetson Nano seems to be promising choice to get all the processing done onboard. © 2020 IEEE.

2020 6th International Conference on Control, Automation and Robotics, ICCAR 2020

Designing of an Amphibian Hexapod with Computer Vision for Rescue Operations

Building and expansion of an efficient transportation network are essential for urban city advancement. However, tracking road development in an area is not an easy task as city planners do not always have access to credible information. A road network mapping framework is proposed which uses a random forest model for pixel-wise road segmentation. Road detection is followed by computer vision post-processing steps including Connected Component Analysis (CCA) and Hough Lines method for network extraction from high-resolution aerial images. The custom dataset used consists of images collected from an urban settlement in India. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

Proceedings of SPIE - The International Society for Optical Engineering

Road network mapping from aerial images

Monitoring tree cover in an area plays an important role in a wide range of applications and advances in UAV technology has made it feasible to capture high resolution imagery which can be used for this purpose. In this study, we adopt a state of the art object detector Mask Region-based CNN (Mask R-CNN1), through transfer learning, for the task of tree segmentation and counting. One bottleneck for the proposed task is the huge amount of data required if the model is required to be scalable to various different geographical regions. Towards this end, we explore the use of a sampling technique based on Gist descriptors and Gabor filtering in order to minimize the amount of training data required for obtaining excellent model performance across images with varied geographical features. This study was conducted across four regions in India, each having a different geographical landscape. We captured a total of 2357 images across all four regions. The final training dataset comprised of 48 images (sampled using the aforementioned method), representative of the entire dataset. Our method demonstrates high quality and scalable tree detection results. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.