PAL Robotics Launches ARAL Humanoid Development Project for Logistics

On November 4, 2026, PAL Robotics introduced ARAL, a humanoid development project for fashion and retail logistics. It aims to create a robot that can work in warehouses designed for people without modifying shelves or existing infrastructure, using the company’s bipedal research platform KANGAROO as its basis. The initial goal is to advance the technology readiness level from TRL 4 to TRL 5.

PAL Robotics has set working without changes to existing shelves or warehouse infrastructure as a goal for ARAL. The project includes developing sensing for objects that cannot be seen visually and actuators that adjust force for handling clothing, setting it apart from automation designed around regular-shaped objects.

ARAL focuses less on the humanoid form itself than on the conditions for deployment—using warehouses without modifying them—and on the perception and manipulation systems needed to handle clothing and hangers. PAL Robotics says it is building on KANGAROO and aiming to advance from TRL 4 to TRL 5. By the company’s description, this places the project in a development phase leading to evaluation in real-world environments, rather than presenting a finished product. Our coverage of PAL Robotics’ robots has included an article introducing TALOS, a humanoid robot for research and development, and an article introducing ARI, a wheeled humanoid intended for social interaction, customer service and research. In contrast, ARAL sets out to develop a robot for specific logistics tasks and warehouse environments. The subject of coverage has shifted from introducing product lines to a project that specifies its work environment and technical challenges. Technically, the design combines tactile and torque sensing to compensate when a camera cannot see into places such as deep boxes, with training in manipulating soft objects such as clothing. The goal of avoiding shelf modifications therefore depends not only on mobility but also on recognizing hard-to-see objects and handling hangers reliably. The plan to train walking and manipulation in simulation, and to teach new tasks using VLA and reinforcement learning, includes a learning stage before physical testing; performance in real-world environments will require separate verification. The plan calls for defining metrics by May 2027, comparing performance with manual work and conducting a controlled pilot at a distribution center. Key points to watch include the specific tasks, the performance metrics and acceptance criteria used for comparison, results in real-world environments, and how the achievement of TRL 5 will be assessed. The current description provides no figures for work speed or throughput, so the project plan alone does not establish its suitability for warehouse operations.

ARAL is intended to combine vision with tactile and torque sensing to handle soft clothing, including inside deep boxes beyond a camera’s view. The plan is to use vision-language-action (VLA) models to understand simple voice commands and train walking and the manipulation of hangers with clothing in simulation. The project also aims to use VLA models and reinforcement learning to teach new tasks. Actuators that detect and adjust force are also part of the development effort. PAL Robotics says it plans to define use cases and metrics by May 2027, establish a test bed at its lab in Barcelona, and develop hardware and software. It plans to compare the robot’s performance with manual work and conduct a controlled pilot at an operating distribution center. ARAL is funded by the European Union’s European Innovation Council (EIC).