cv
Ph.D. student in Mechanical Engineering and M.Sc. student in Computer Science at USC. Research focus on vision-language-action models, robotic task planning, and agentic AI.
Basics
| Name | Ardalan Aryashad |
| Label | Ph.D. Student in Mechanical Engineering & M.Sc. Student in Computer Science |
| aryashad@usc.edu | |
| Phone | (213) 551-5016 |
| Url | https://ardal-on.github.io |
| Summary | Ph.D. student at the USC IMPACT Lab working on vision-language-action models, robotic task planning, and agentic AI for model-based systems engineering. |
Education
-
2026.01 - Present Los Angeles, CA
-
2023.08 - Present Los Angeles, CA
Ph.D.
University of Southern California
Mechanical Engineering
- Vision-Language-Action Models
- Robotic Task Planning
- Agentic AI for MBSE
-
2018.09 - 2023.08 Tehran, Iran
Work
-
2023.08 - Present Research Assistant
IMPACT Laboratory, University of Southern California
Research on vision-language-action models and agentic AI for mechanical systems engineering.
- Developed a graph-based planning framework (VLAG) for VLA models in long-horizon manipulation tasks.
- Trained a CLIP-based vision model achieving 95% accuracy on environmental key state prediction.
- Fine-tuned a sentence transformer model reaching 98% instruction-to-task matching accuracy.
- Trained a Transformer-based action model validated on the CALVIN dataset.
- Designed single and multi-agent workflows for MBSE modeler tasks.
- Created a Visual Paradigm plugin to auto-generate SysML diagrams using a custom LLM agent.
- Implemented a text extraction pipeline integrated into a RAG system for an AI agent.
-
2022.08 - 2023.06 Research Assistant
Micro-Nano Systems Lab, Sharif University of Technology
Control system design for a PCR microsystem.
- Conducted system identification experiments on a PCR microsystem.
- Designed a 2-DOF PID controller reducing settling time by 17% and eliminating overshoot.
- Trained a neural network predictive model and implemented it in MPC.
- Executed MPC on a Raspberry Pi using MATLAB/Simulink.
-
2021.09 - 2022.06 Research Assistant
Experimental Solid Mechanics and Manufacturing Lab, Sharif University of Technology
Pipe material selection and surge pressure simulation.
- Reviewed literature on essential parameters for pipe material selection in water networks.
- Conducted surveys and interviews with design engineers to rank design parameters.
- Simulated surge pressure phenomena using FEM in Iranian PE pipes.
-
2021.06 - 2021.08 Student Intern
Automotive Industries Research & Innovation Center of SAIPA (AIRIC)
Regenerative braking simulation and analysis.
- Researched and simulated regenerative braking performance for Saipa Shahin.
- Developed a script to convert Google Maps data into simulation-compatible road profiles.
- Simulated regenerative braking across multiple real-world routes.
-
2020.06 - 2020.08 Student Intern
Industrial Saipa Heavy Dies Manufacturing
Maintenance department support and process automation.
- Developed an Excel-based preventive maintenance schedule.
- Built a web-based maintenance request system to improve department workflow.
Volunteer
-
2024.01 - Present Mentor
IMPACT Laboratory, University of Southern California
Mentoring students on RAG modules, hyperparameter optimization for robotic manipulation, and dataset collection.
-
2024.01 - Present Peer Reviewer
IDETC-CIE, WACV, International Journal of Computer Integrated Manufacturing
Reviewing manuscripts in robotics, computer vision, and mechanical engineering.
-
2020.09 - 2023.06 Teaching Assistant
Mechanical Engineering Department, Sharif University of Technology
Applied Finite Element, Control Design, Mechanical Design, Embedded Systems, Dynamics, and Statics courses.
Awards
- 2023.08.01
Andrew and Erna Viterbi Graduate Student Fellowship
Department of Aerospace & Mechanical Engineering, USC
Fellowship awarded to incoming Ph.D. students in the AME department.
- 2022.06.01
Best Teaching Assistant in Mechanical Engineering
Sharif University of Technology
Voted by students across the Mechanical Engineering department.
- 2019.01.01
National University Entrance Exam — Ranked 146th out of 144,000
Iran National Organization for Educational Testing
Mathematics branch, national rank 146 out of 144,000 participants.
Publications
-
2025.01.01 VLAG: Graph-Based Planning for Vision-Language-Action Models in Long Horizon Manipulation Tasks
ASME IDETC/CIE 2025
A modular graph-based planning framework for VLA models that decomposes long-horizon tasks into structured subtasks using dedicated vision, language, and action modules.
-
2024.01.01 From Filters to VLMs: Benchmarking Defogging Methods through Object Detection and Segmentation Performance
Workshop on Image/Video/Audio Quality Assessment, WACV 2026
A benchmark pipeline evaluating fog-removal methods by their effect on downstream object detection and segmentation tasks across synthetic and real-world fog datasets.
Skills
| Software | |
| Matlab | |
| Simulink | |
| Simscape | |
| SolidWorks | |
| Abaqus | |
| NX | |
| Ansys | |
| Proteus | |
| Cursor |
| Programming Languages & Tools | |
| Python | |
| PyTorch | |
| MuJoCo | |
| PyBullet | |
| lerobot | |
| C/C++ | |
| Linux | |
| Git | |
| AWS | |
| Slurm |
| Prototyping & Hardware | |
| Arduino | |
| Raspberry Pi | |
| 3D Printing | |
| Lathe | |
| Soldering | |
| Serial Communication | |
| Servo/Stepper Motor Control |
Languages
| Persian | |
| Native speaker |
| English | |
| Fluent |
Interests
| Research | |
| Vision-Language-Action Models | |
| Robotic Task Planning | |
| Agentic AI | |
| Model-Based Systems Engineering | |
| Mechanical Design Automation |
References
| Prof. Yan Jin | |
| Advisor, IMPACT Laboratory, University of Southern California. Available upon request. |
Projects
- 2024.01 - 2025.01
VLAG
Graph-based planning framework for vision-language-action models in long-horizon robotic manipulation.
- IDETC/CIE 2025
- USC IMPACT Lab
- 2023.01 - 2024.01
Fog Removal Benchmark
Benchmark pipeline evaluating defogging methods on downstream object detection and segmentation tasks.
- WACV 2026 Workshop
- 54 pipelines evaluated
- 2024.01 - Present
SEMAA
Single and multi-agent assistant workflows for mechanical systems engineering and MBSE modeler tasks.
- USC IMPACT Lab
- semaa.site
- 2023.08 - 2024.06
USC Formula Student Racing
Suspension design and DAQ system for USC's FSAE competition entry. Top 10 Design Finals at FSAE Michigan 2024.
- Top 10 Design FSAE Michigan 2024
- Suspension optimization
- Raspberry Pi NAS for MoTeC ECU logs