Research
Research themes
Our research spans design methodology, AI-integrated engineering, autonomous systems, multiagent self-organizing systems, VLA robots, and organizational approaches to collaborative engineering.
Over the years, USC IMPACT LAB has established a broad portfolio spanning design methodology, AI-integrated systems, self-organizing systems, VLA robotic models, autonomous systems, and organization modeling, with a focus on AI-integrated engineering, where People + AI are at the core of engineering.
AI-Integrated
AI-integrated engineering systems
We develop AI-integrated engineering systems by embedding agentic AI technologies throughout the systems engineering lifecycle to enable more connected, intelligent, and efficient engineering processes.
Our research aims to smooth the engineering digital thread by transforming fragmented islands of information into coordinated, interoperable engineering ecosystems. Core research areas include AI-supported SysML modeling, lifecycle data backbones, toolchain integration, engineering knowledge grounding, and efficient multi-agent orchestration.
Application domains include shipbuilding design and construction as well as electrical systems engineering, where increasingly complex workflows require tighter integration between engineering knowledge, digital infrastructure, and AI-enabled decision support.
Self-Organizing
Self-organizing systems
In self-organizing systems, we explore the dynamics of complex systems, particularly how agents can learn and adapt to task environments without centralized control. Our early work explored field-based approaches to behavior regulation.
Recent studies provide insights into social learning and reward-shaping mechanisms in multiagent systems, offering approaches to foster collaboration and knowledge sharing in decentralized environments. This research demonstrates the potential of self-organizing systems to solve complex engineering problems through emergent behavior and decentralized decision-making.
VLA Robots
VLA model-based robots
Our research on VLA model-based robots aims to realize AI-integrated engineering in manufacturing jobshops, where humans and AI systems work collaboratively. Robots are expected to perceive their environments, communicate naturally with human operators, and execute actions based on both situational context and human instructions.
The research addresses fundamental challenges in vision-language-action modeling, multimodal learning, reasoning, planning, and real-time robotic computing. The VLAG framework, Vision-Language-Action with Graph Routing, investigates modular planning for long-horizon robotic manipulation, integrating perception, language understanding, graph-based task planning, and specialized action experts.
Methodology
Design methods
IMPACT LAB's research on design methodology advances our understanding of how design processes can be augmented through human-AI interactions that leverage insights into cognitive design behaviors. We develop research methods and design processes that facilitate the creation of innovative designs.
By integrating AI and machine learning techniques, we aim to develop intelligent design support systems that enhance the creativity and efficiency of design processes. Recent work on ChatGPT and finetuned BERT demonstrates how AI technologies can support intelligent design systems, while studies on design knowledge extraction and data-enabled sketch search show how data can improve design decision-making and ideation.
Autonomy
Autonomous systems
Our research enhances the safety, reliability, and efficiency of autonomous navigation and operations. Work on ship collision avoidance and path planning applies advanced AI methods to critical challenges in autonomous systems.
Our long-range risk-aware path planning research introduces approaches that account for long-range risks and integrate sophisticated risk assessment models into autonomous navigation. This work supports safer and more efficient autonomous vehicles and vessels, including a four-year autonomous ships project sponsored by the Japan Consortium of Autonomous Ships.
Project video: https://youtu.be/tCNArfiYClo.
Organization
Organizational and collaborative engineering
Recognizing the importance of teamwork and collaboration in engineering projects, we study organizational and collaborative engineering. Early work pioneered computational organization modeling and contributed to the development of the Virtual Design Team model.
Our work examines how simulation-based process design, negotiation strategies, and team dynamics influence the success of collaborative engineering efforts. This research highlights the role of communication, negotiation, and effective team management in achieving project objectives and fostering innovation in engineering design.