Artificial Intelligence

sim1

Screenshot-from-2016-06-15-10_53_02

Screenshot-from-2016-06-09-12_07_31

New Slide

New Slide

SceneFlow

segment

Map

Lanes


In the Artificial Intelligence Division, we build the mind of the entire system. We are tightly intertwined with the Sensing and Automation Division. On obtaining real world data from the S&A Division, we are responsible for two fundamental tasks: to perceive our environment and hence make decisions accordingly and are thus divided into two subdivisions:-
PERCEPTION
1. Detection and localization of static and dynamic obstacles
2. Classification of different obstacles like pedestrian, vehicles, trees, buildings, traffic lights and signs in real time#
3. Segmentation of road into different permissible and non-permissible lanes
4. 3D reconstruction of data from cameras and LIDARs
5. Scene flow
6. Predicting depth using mono camera
PLANNING
1. Semi-autonomous mobile real robots
2. Efficient, probalistic mapping
3. Layered occupany grid of the environment using multiple sensors
4. Complex path planning, intelligent decision making with optimal routes, control algorithms to avoid obstacles and providing the smoothest experience
5. Prioritized navigation waypoint system
HIGH PERFORMANCE COMPUTING
1. Parallel computing using CUDA for GPU acceleration
2. Multi-GPU cluster for training deep learning models
SIMULATION
1. Fully autonomous vehicles in simulation
2. Accurate model of Mahindra e2o car