Multibody dynamics simulation plays a vital role in many fields such as robotics, industry, aerospace, and automotive engineering. In such applications, the development of efficient solvers for contact and friction is crucial. This thesis focuses on an Accelerated Projected Gradient Descent (APGD) method inspired by Nesterov’s acceleration scheme. At each simulation timestep, the method formulates the computation of normal and frictional contact forces as a convex quadratic program with conic constraints, and efficiently solves it using a momentum-accelerated projection algorithm. Compared to the traditional Gauss-Seidel method, APGD can reduce simulation time by one to two orders of magnitude for large-scale rigid body systems with frictional contact. Moreover, unlike Gauss-Seidel, the APGD approach is highly parallelizable and well-suited for simulations involving millions of interacting bodies.
The objective of this thesis is to implement the APGD method within a multibody dynamics simulation framework to enhance computational performance for contact-rich scenarios. The workflow begins by reformulating the Non-Smooth Contact (NSC) or Cone Complementarity Problem (CCP) into an equivalent convex quadratic program. The APGD solver is then integrated into a C++-based simulation engine. Finally, the performance of the proposed method is benchmarked against the traditional Gauss-Seidel method to evaluate improvements in convergence speed and computational efficiency.
Keywords: Multibody dynamics simulation, optimization, friction, contact
Requirements:
-
You are studying Electrical Engineering, Automation, or Robotics Systems Engineering.
-
You are interested in robot simulation or game physics engines; ideally, you have studied multibody dynamics or robotics dynamics.
- Ideally, you have programming skills in C++
Betreuer: Shao, Email:
