I'm a robotics engineer designing robotics and control solutions using conventional and learning-based control techniques.
My work primarily has focused on:
- Motion planning for robotic systems
- Learning-based control
- Optimization-based control
- Systems modeling and analysis
This project focuses on solving the motion planning for multiple magnetic robots under a uniform magnetic field. It consists of three subprojects addressing different challenges related to the problem. The multi-robot system functions in several selectable modes and is a hybrid system. During operation, the robots must avoid collisions with other robots and environmental obstacles.
This part addresses motion planning while accounting for modeling uncertainties. A tube-based model predictive control is adapted for the hybrid system to reduce the negative impacts of model uncertainties. The collision avoidance constraints are elegantly derived, leading to a Quadratically Constrained Quadratic Program (QCQP) that is linear in uncertain variables, simplifying the implementation of constraint tightening. This approach improves the success rate of executing motion plans.
This part focuses on navigating robots through obstacles. The RRT* algorithm is tailored to accommodate the hybrid and multi-modal nature of the multi-robot system.
More information can be found .
This part represents the initial study of this project. The multi-robot system is analyzed to identify the requirements for controllability, and the motion planning problem is formulated as an optimization problem. The collision avoidance between robots is achieved through nonlinear nonconvex constraints from straightforward geometrical analysis.
More information can be found .
This project aims to enhance the SLM metal 3D printing process by controlling the workpiece temperature. I developed a Gaussian Process-based learning model predictive controller to optimize laser power and speed throughout the process to reduce defects.
Moreinformation can be found .
The goal of this project was to create an agile autonomous vehicle testbed to develop learning-based control techniques. I modified a Traxxas RC truck by adding sensors, a microcontroller, and an NVIDIA TX2 board, and then implemented algorithms for basic path-following at a constant speed.
More information can be found in .
This was the term project for an optimization class. I designed a reinforcement learning-based tracking controller based on double Q-learning for a 2D robotic manipulator.
This project shows that an Approximate Dynamic Programming solution of the nonlinear optimal tracking control of affine nonlinear systems is stable when the reference trajectory in the training data is sampled from a uniform distribution. I provided a mathematical proof for this stability and implemented the approach on a Delta parallel robot built myself.
More information can be found .
📫 My Contact
- 📧 Email: farshidasadi47@yahoo.com
- 🔗 LinkedIn: linkedin.com/in/farshid-asadi