Description

We will be working alongside Embedded Systems at Purdue to bring WALL-E to life. While they handle the engineering aspect that makes WALL-E function, we will build WALL-E’s intelligence.

The end goal: To create a WALL-E that faithfully adapts his namesake. WALL-E should:

• Have personality
• Be able to engage in some conversational capacity with humans
• Find objects that he cannot directly see
• Pathfind through his surroundings
• Autonomously combine his available skills to complete complex tasks

We will use Natural Language Processing (NLP) to create Behavior trees, allowing WALL-E to turn spoken commands into actionable objectives, and react accordingly under unforeseen circumstances. We will also build off of previous literature and create new architectures using foundational blocks like transformers, CNNs, VLMs, and more to achieve strong and lightweight autonomous navigation and spatially aware object search. Finally, we will work with a team of strong-minded mechanical and electrical engineers to take our intelligence from lines of code stuck on the cloud to influencing the actions of a physical, tangible WALL-E robot.

Prerequisites

• Basic ML Python libraries (NumPy, PyTorch, Tensorflow, etc.)
• Understanding of foundational ML concepts (Backpropagation, MLP, CNN, Transformer, etc.)