We are developing a system to make autonomous deliveries on the UT campus.
An interdisciplinary team spanning across four UT labs is developing a building-to-building delivery system for the UT Austin campus using robots that will improve contactless deliveries and save people time. These robots will cross complex terrain, navigating around people, cars, and other obstacles typical to campus roadways. An iOS app will allow people to place, track and securely retrieve packages. From offices to student dormitories, these deliveries will improve safety and efficiency on the UT Austin campus and could be used to limit exposure during the COVID-19 pandemic.
The initial project was funded by UT Good Systems and led by Dr. Junfeng Jiao, Dr. Luis Sentis, Dr. Joydeep Biswas, Dr. Justin Hart.
Making autonomous deliveries requires coordinating many components.
Our robots have a sense of where they are.
These robots will autonomously navigate around campus. To make this happen, a robot must be able to understand its current location and its surrounding environment. We accomplish this task by analyzing point cloud data collected from a lidar sensor and depth cameras. State of the art software identifies landmarks in this point cloud data and compares these identified features to a known map of the area to localize the robot.
Our robots plan for unexpected obstacles.
Once a robot knows its location, it needs to plan a path to its destination and continuously adjust its movements to match that plan. At the same time, the robot must handle unexpected obstacles like cars, people and bikes in its path. A robust graph navigation planner undertakes these tasks, allowing robots to operate in a dynamic human environment.
It all comes together in the architecture.
There are many independent software services working together to make autonomous robot deliveries possible. A sophisticated software architecture works to connect these components, ensuring that customer orders created in the App reach the robot for delivery and status updates reach the customer at every step without human intervention.
Our App keeps users updated at every step.
The App will allow customers to see the current location of their package and watch the movements of the delivery robot on a map. Once the robot arrives,the App will notify the customer that it is time to retrieve their package and send the customer a code to access their order.
Order a free lemonade on campus.
To test our system, we plan to deliver free lemonades on campus to UT-affiliated persons. Customers will be able order a lemonade in the App, and have it delivered to a select number of buildings. After this proof of concept, we hope to see our system integrate with UT dining services and other restaurants on campus.