Bowen Feng

Automated 3D Modeling

The goal of our project is to create a 3D reconstruction model of an arbitrarily object. We aimed to design an automated 3D reconstruction system to make people without any CAD modeling experience can construct 3D detailed model from random objects.

Autonomous driving with deep Q learning in Carla Urban Simulator

The goal of this independent research is to create a deep Q learning model for a vehicle in Carla Urban Simulator. The target pipeline is to make spawned vehicle learn the environment and drive itself without collide under no traffic or line following rules.

Object avoidance and line following with Jackal robot using CNN and Neural Circuit Policy

Neuron Circuit policy gives us a more intelligent way to use less data and fewer layers during the learning while finishing the task with more robustness. We design the experiment to ask Jackal robot doing autonomous navigation tasks to demonstrate using neuron circuit policy with fewer control neurons can still result in a satisfactory performance.

Use robot to bend flexible wire to specific shape as tools with optimal control and active learning

Our goal in general is to construct a systematic method for using a soft, bendable rod as a tool. Consider a soft wire with fixed length held by two robot grippers, related work done by Timothy Bretl's lab proved that any curve traced by this wire when in static equilibrium is a local solution to a geometric optimal control problem. Based on this framework, the shape of a flexible wire can be predicted given the specification of the two endpoints of the wire.