Video Friday: Digit Learns to Deadlift

2026-04-17 · Source: IEEE Spectrum · Field: Technology & Digital — Robotics & Autonomous Systems, Artificial Intelligence & Machine Learning · Depth: Novice, short

Summary

This "Video Friday" compilation from IEEE Spectrum Robotics showcases recent advancements and applications in robotics through various video demonstrations. Highlights include Agility Robotics' Digit performing a 65-pound deadlift, demonstrating enhanced whole-body coordination and actuator resilience, and Gatlin Robotics' debut commercial for its Robot-as-a-Service (RaaS) contract. Harvard researchers present RAnts, a swarm of simple ant-like robots capable of collective excavation and construction without central control, adaptable by tuning two parameters. Michigan Robotics unveils a microcombustion actuator that challenges assumptions about soft robotics by demonstrating fast, forceful, and explosive motion. Other notable entries feature Unitree Robotics' H1 humanoid running at high speed, LimX Dynamics' TRON 1 AI photography robot navigating complex terrains, and MagicLab's large-scale deployment of robot dogs and humanoids for multiagent control system validation. The compilation also lists upcoming robotics events like ICRA 2026, RSS 2026, and a Summer School on Multi-Robot Systems.

Key takeaway

For research scientists developing advanced robotic systems, this compilation underscores the importance of integrating robust simulation for dynamic tasks and exploring emergent behaviors in multi-agent systems. Your work on whole-body coordination and actuator resilience, as seen with Agility's Digit, is critical for real-world applicability. Additionally, investigate how tuning simple parameters in swarm robotics, like Harvard's RAnts, can yield adaptive group behaviors, potentially simplifying complex control challenges.

Key insights

Robotics advancements span humanoids, swarms, and soft actuators, pushing boundaries in coordination, autonomy, and real-world deployment.

Principles

• Simple agents can yield complex group behavior.
• Soft actuators can be fast and forceful.
• Simulation accounts for load distribution and grip forces.

Method

Training policies for robots like Digit involves simulating object interaction to account for load distribution, grip forces, and center of mass changes, ensuring dynamically balanced real-world performance.

In practice

• Use simulation to refine robot load handling.
• Explore multi-agent systems for complex tasks.
• Consider soft robotics for high-speed applications.

Topics

• Digit Robot
• Swarm Robotics
• Soft Actuators
• Robot-as-a-Service
• Embodied AI

Best for: Research Scientist, Robotics Engineer, AI Scientist, AI Student

Related on AIssential

• Video Friday: Digit Learns to Dance—Virtually Overnight — Advanced robotics are achieving new performance thresholds in control, learning, and human interaction.
• A Mechanistic Model for Collective Motion from Sensorimotor Regularities — Collective behavior emerges from individual sensorimotor regularities, not abstract interaction forces.
• Video Friday: Who Wins in Robot vs. Pro Ping-Pong Player? — Robotics research is rapidly advancing high-speed autonomous systems, versatile locomotion, and efficient actuation.
• A neural blueprint for human-like intelligence in soft robots — A neural-inspired AI control system enables soft robots to learn once and adapt instantly to diverse tasks and disturbances.
• AgentJet: A Flexible Swarm Training Framework for Agentic Reinforcement Learning — AgentJet decouples LLM agent rollouts from model optimization for flexible, fault-tolerant, and efficient distributed reinforcement learning.
• Why Do Humanoid Robots Still Struggle With the Small Stuff? — Despite AI advancements, humanoid robots still struggle with fundamental physics like force control for reliable real-world interaction.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by IEEE Spectrum.