Mechanically Intelligent Soft Robotics 

 Abstract

Traditional rigid robotics is largely built upon a controller-centric architecture, fundamentally separating sensing, computation, and actuation into distinct components. However, soft robots approach unstructured environments differently; they excel at safe, adaptive interactions by embedding intelligence directly into the physical body of the robot itself. By transitioning away from siloed electronic components, these systems simplify the overall system architecture while simultaneously driving improvements in robustness, adaptability, and autonomy.

We define Mechanical Intelligence in soft robots as enabling soft robots to sense, adapt, and make decisions by leveraging their inherent physical structure and mechanics. By embedding sensing, regulation, and decision-like behaviors directly into physical interactions, and transforming environmental inputs into coordinated motion and state transitions, mechanical intelligence minimizes reliance on computationally heavy processing and complex feedback loops. This paradigm is rapidly advancing through recent developments in bio-inspired design, fluidic oscillators, mono-/bi-/multi-stability, and mechanical logic gates.

This workshop brings together the soft robotics community with experts in mechanics, structures, and design to explore the emerging field of mechanically intelligent soft robotics. We will highlight state-of-the-art advancements, pinpoint grand challenges and evaluation metrics, and forge cross-disciplinary pathways. Ultimately, the workshop seeks to catalyze new interdisciplinary collaborations and advance the development of robotic systems that integrate mechanical intelligence.

 Content

The primary objective of this workshop is to investigate how mechanical intelligence embedded in the body of soft robots can enable sensing, adaptation, and decision-making during physical interactions with unstructured environments. This workshop aims to bring together experts in robotics, mechanics, soft materials, and design to define a shared language for mechanically intelligent soft robots, unify future research directions.

Reflecting the State-of-the-Art: We are inviting leading researchers from soft robotics as well as experts from complementary fields such as mechanics, materials science, and design to exchange perspectives on cutting-edge developments and future trajectories. To ensure fresh ideas and highlight emerging talent, these invited talks will be joined by high-quality lightning talks from students and postdocs, rigorously selected from a competitive abstract submission process. Engaging Attendees and Exchanging Ideas: The workshop format is intentionally designed to prioritize interactivity over passive listening. Ample time will be reserved for rigorous Q&A following each presentation. Furthermore, we will host dynamic poster and live physical demonstration sessions during extended coffee breaks to encourage hands-on engagement. The event will culminate in a dedicated, interactive panel discussion to synthesize the day's ideas. Expanding Originality and Diversity at IROS 2026: This workshop explicitly expands the originality and diversity of IROS content, with the focus on exploring how to utilize mechanical intelligence to enhance the dynamic performance of soft robot systems and simplify their complex designs. By spotlighting "mechanical intelligence", we are introducing a rapidly emerging paradigm to the broader conversation and providing a dedicated, high-visibility venue for this community to disseminate groundbreaking results.

 Organizer team

Contact: 

Corresponding: Haitao Qing, haitao.qing@berkeley.edu