3D Perception and Prediction for Autonomous Driving

For an Autonomous Vehicle (AV) to make decisions and drive independently on urban streets, the problem at hand can be broken down into many phases, two of which are perception and prediction. Perception refers to the process of extracting valuable information from the environment using data collected by sensors such as LIDAR and camera. This includes detection of cars, ped estrians, lanes among many objects. Prediction refers to the process of tracking all the known objects and predicting the possible future actions so as to enable the autonomous vehicle to make informed decisions. Traditionally these tasks are done sequentially and independently one after another. This makes uncertainty hard to propagate from perception to prediction. The aim of this project is to build a deep learning model that does combined 3D perception and prediction.

Faculty Supervisor:

Sanja Fidler

Student:

Satya Krishna Gorti

Partner:

Uber Advanced Technologies Group

Discipline:

Computer science

Sector:

Information and communications technologies

University:

Program: