Robots and Embodied Agents in Collaborative Construction
Future of Construction 2026 Workshop
R2-D2 and C-3PO, the robot characters from Star Wars (1977), make it easy to imagine an intelligent agent with a body. Her (2013) gives the opposite image: a voice with no body, but a strong personality and a memorable relationship with its user.
Sci-fi offers concrete examples for a question: when an AI agent is connected to a construction robot, where should people feel the agent is embodied? In this workshop, we worked with participants to tailor different aspects of embodiment and speculate on how that can change collaboration among people, robots, and AI agents.
The workshop was built on the MAVE system but used a simplified one-day setup. Instead of a full multi-agent architecture, participants worked with a single LLM agent connected to AR interaction and robotic task logic.
The student projects showed how quickly embodiment becomes a question of relationship. Some groups imagined trust growing through conversation and personalisation. Others focused on prediction, validation, or shared intelligence across many machines.
Future of Construction 2026 Workshop
R2-D2 and C-3PO, the robot characters from Star Wars (1977), make it easy to imagine an intelligent agent with a body. Her (2013) gives the opposite image: a voice with no body, but a strong personality and a memorable relationship with its user.
Sci-fi offers concrete examples for a question: when an AI agent is connected to a construction robot, where should people feel the agent is embodied? In this workshop, we worked with participants to tailor different aspects of embodiment and speculate on how that can change collaboration among people, robots, and AI agents.
The workshop was built on the MAVE system but used a simplified one-day setup. Instead of a full multi-agent architecture, participants worked with a single LLM agent connected to AR interaction and robotic task logic.
The student projects showed how quickly embodiment becomes a question of relationship. Some groups imagined trust growing through conversation and personalisation. Others focused on prediction, validation, or shared intelligence across many machines.
Instructors
Xiliu Yang, Samuel Slezák
Participants
Pierpaolo Ruttico, Jiawei Rong, Oliver Englhardt, Anton Savov, Wenqian Yang, Liana O’Cleirigh, Raphael Wolfensberger, Michael Mangold, Fabi Braun, Marco Galli, Clemens Loidolt, Xindi Zhang, Chenjie Hu
Pierpaolo Ruttico, Jiawei Rong, Oliver Englhardt, Anton Savov, Wenqian Yang, Liana O’Cleirigh, Raphael Wolfensberger, Michael Mangold, Fabi Braun, Marco Galli, Clemens Loidolt, Xindi Zhang, Chenjie Hu
Financal Support and Sponsoring
Future of Construction 2026, ETH Zurich
Future of Construction 2026, ETH Zurich