Robots relieve warehouse staff during unloading
MIT start-up combines AI, sensor technology and robotics for robust logistics automation
The MIT start-up Pickle Robot Company is developing AI-controlled unloading robots that independently lift cartons from trucks and place them on conveyor belts.
The Pickle Robot Company develops single-armed robots that autonomously unload trailers and place boxes weighing up to 50 pounds on conveyor belts. This takes over repetitive, physically demanding work that leads to high injury rates in warehouses. The robot platform combines generative AI, classic machine learning models, sensor technology and cameras, enabling it to immediately orient itself in new environments, recognise objects and dynamically adapt its movements.
One focus is on perception: pre-trained AI models are adapted to the lighting, material and geometric conditions in warehouses. Supplementary models control gripping point selection, path planning and error detection. KUKA industrial robot arms are mounted on mobile bases with their own computing units, enabling the robots to adjust their position autonomously. Depending on the package size, they can move 400 to 1,500 boxes per hour.
Tests in extreme environments
The founding team – all three alumni of MIT in the USA – identified the problem after empirically examining logistics processes and documenting high employee turnover. Initial prototypes only worked reliably for a short time, but aroused great interest in the industry and enabled new financing.
At the same time, a software platform is being developed that networks various robot systems such as autonomous forklifts and palletisers to efficiently coordinate material flows.
Pickle sees unloading trucks as the first step in a platform strategy: in the future, robots will take on other repetitive tasks in logistics, production and retail, thereby specifically reducing the workload on humans.
