Inspired by the cooperation of cells in tissues, researchers have developed a robotic collective system capable of transitioning between rigid and solid structures that can also support hundreds of times its own weight. The advancement overcomes a core challenge in the development of so-called “robotic materials” – cohesive networks of individual robotic units that function as a single dynamic, adaptive structure. Realizing these systems presents a fundamental challenge: this “material” must at once be strong and stiff enough to support loads, yet at the flip of a switch, be able to flow freely to take new forms. Unlike inert materials and conventional robotic systems, living embryonic tissues possess the remarkable ability to internally regulate their mechanical properties across space and time. Drawing inspiration from living embryonic tissues, which regulate mechanical properties through coordinated cellular behavior, Matthew Devlin and colleagues engineered robotic collectives that mimic key cell-cell interaction mechanisms using motorized gears, photoreceptors, and rolling magnets. These features enable precise control over force fluctuations and polarity, allowing the system to dynamically adjust rigidity and fluidity. Devlin et al. demonstrated structure formation, with robotic units forming pillars that merged into a stable, load-bearing arch. The collective also exhibited self-healing, fluidizing to close structural defects. And it exhibited object manipulation, applying directed forces to move items. Additionally, the system adapted into functional tools, flowing around objects before rigidifying into a wrench capable of exerting torque. The researchers further showcased supporting structures, where the collective bore loads exceeded individual unit weight – supporting a human (~700 Newtons) before effortlessly transitioning back into a fluid state.

High-performance computing (HPC) systems – advanced computing ensembles that harness deliver massive processing power – are used for a range of applications, and the demand for them has increased with the rise of generative artificial intelligence (AI). However, for both traditional uses and to advance the power of AI, technical advances in HPC are greatly needed, say Ewa Deelman and colleagues in a Policy Forum. “With international competition for leadership in computing intensifying, without a renewed commitment, we fear that the US will soon lose scientific computing leadership and technological independence,” say the authors. They outline how HPC systems are in a state of transition, shaped by both technology constraints and market forces. Among the major challenges for HPC systems are the wide use of chips with lower-precision arithmetic – insufficient for applications important for science, engineering, and defense – as well as the power consumption of current machines. Deelman and colleagues highlight how over time, the global HPC landscape has evolved, placing HPC in a geopolitical arena where nations compete for technological sovereignty. They provide examples of HPC-related initiatives in different nations, which “reveal deep-seated policy and technical tensions around national security, international collaboration, and market independence.” The authors write, “With this Policy Forum, we aim to bring attention to the totality of challenges and opportunities in HPC and advocate for a multiagency, ‘whole-nation,’ and internationally collaborative effort to reenergize HPC R&D.”