Agility Robotics Launches Digit 5 with Uncaged Safety Architecture for Logistics

Agility Robotics has officially unveiled Digit 5, the latest iteration of its humanoid robot engineered specifically for high-throughput logistics and industrial automation. Drawing from more than 65,000 operational hours logged by Digit 4 across enterprise sites like GXO and Amazon, Digit 5 introduces a 40% increase in payload capacity to 50 pounds, a fast-charging battery architecture capable of a full recharge in 9 minutes for 90 minutes of continuous runtime, and ISO-standard tool flanges for swappable end-effectors. Crucially, the platform incorporates a dedicated safety controller built with NVIDIA IGX Thor and the Halos safety framework, enabling the robot to work directly alongside humans in shared facility spaces without perimeter fencing. The introduction of an uncaged functional safety model fundamentally changes the operational economics of physical automation for enterprise logistics teams. Historically, deploying heavy autonomous machinery required dedicated work cells, light curtains, or physical barriers to satisfy industrial safety regulations, adding real estate overhead and preventing seamless handoffs between human pickers and robotic carriers. By embedding an independent safety compute stack that governs dynamic human detection and real-time motion cues, operations teams can deploy bipedal humanoids directly into existing brownfield facilities without overhauling warehouse layouts. This release highlights the broader acceleration of edge physical AI converging with deterministic functional safety standards. Previously, high-level perception models and deterministic motion-control safety ran in isolated, disparate compute silos. The integration of high-throughput AI silicon like NVIDIA IGX Thor alongside certified safety layers allows vision-based spatial intelligence to inform safety-rated motion controllers directly. This hybrid design ensures the robot maintains compliant emergency-stop and obstacle-avoidance guarantees while retaining the dynamic adaptability needed to manipulate unstructured parcels. For systems architects and DevOps practitioners managing robotics fleets, the near-term task is modernizing orchestration pipelines to support heterogenous mobile agents. Operating uncaged humanoids requires integrating high-frequency spatial telemetry and continuous health diagnostics into existing Warehouse Management Systems (WMS) and Manufacturing Execution Systems (MES). Engineering teams should begin auditing their industrial edge networks for low-latency communication and adopting simulation-to-real testing pipelines to validate fleet-level safety constraints before enterprise rollouts begin in 2027.