evoBOT®: Autonomous transport robot for logistics

evoBOT® represents the new generation of autonomous mobile robot systems that combine highly dynamic balanced stability with high logistics performance.

The project investigates how a transport robot based on the principle of an inverted pendulum can perform a wide range of tasks, from load carrier transport to human-robot collaboration. Simulation-based development supported by AI and the resulting digital twins shorten development times and create a robust foundation for test beds in the implementation of concrete practical applications.

The industry faces high throughput, a shortage of skilled workers, and increasing pressure to automate. In warehouses, production areas, and freight terminals, employees move heavy loads and handle hazardous goods.

The evoBOT® project is developing a modular, highly dynamic transport robot on two wheels as a new class of autonomous mobile robot systems. Researchers at Fraunhofer IML are developing a prototype that combines key manipulation functions such as pushing, pulling, turning, and handing over, as well as the transport of objects, in a single system.

The system’s modular design allows it to tackle a wide variety of intralogistics processes. Thanks to its significantly simpler design with two wheels and fewer actuators, it is more efficient and cost-effective than, for example, a humanoid robot. This makes it possible for a single system to perform tasks that previously required multiple specialized AMR systems.

Contact our institute to discuss collaborations on evoBOT®.

• Use of simulation environments, AI, and motion capture to create cyber-physical twins.
• Testing of evoBOT® mobile robotics in laboratory environments.
• Optimization of intralogistics processes, from individual load carrier transport to the handling of hazardous materials.

Send a cooperation request regarding evoBOT®

Send a cooperation request regarding evoBOT®

"With the development of the evoBOT, we are addressing performance gaps in existing robotic systems in industrial settings, such as the simultaneous transport and manipulation of load carriers and goods. We are addressing autonomous use cases as well as human-machine interaction. In this context, we are lowering the barrier to interaction through an optimized design."
Leon Siebel-Achenbach, MBA, conducts research on mobile robotics at Fraunhofer IML

The technical solution utilizes the principle of the inverted pendulum without an external counterweight and balances itself. The evoBOT® navigates ramps, edges, and uneven pavement outdoors while remaining stable at speeds of up to 10 m/s.

The modular platform integrates two gripper arms that can pick up objects such as boxes, packages, or hazardous goods directly from the ground or a conveyor and set them down at variable heights. Thanks to interchangeable grippers, the system is suitable for different load carriers. Employees use the robot as an assistant, for example, to hand over loads, transport packages over longer distances, or to reduce the physical strain of lifting and overhead work in the warehouse.

AI support is used for simulation-based development. First, a simulated model of evoBOT® is generated and transferred to the physical robot. Researchers examine the robot’s behavior, compare it with the simulation, and optimize the model based on this. evoBOT® serves as both a digital twin and a development platform for software and sensor components.

Use evoBOT® as a reference for our AMR expertise.

• From concept to prototypical implementation of Autonomous Mobile Robots based on our work on evoBOT®.
• Modular, highly dynamic two-wheeled transport robot system with an inverted pendulum principle as a reference for AMR platform designs.
• Simulation-based development supported by AI and the resulting digital twins shorten the development times for your mobile robot fleets.
• Collaborative research projects on collaborative robots in logistics, industry, and complex urban environments.

Schedule a meeting about AMR collaborations

In a warehouse, evoBOT® performs a wide range of transport and handling tasks that previously had to be covered by various AMR robots or industrial trucks. The goal is to relieve employees of physically demanding tasks and reduce process times.

The modular, highly dynamic transport robot picks up boxes and packages from the floor or a conveyor, transports them across various surfaces, and sets them down at variable heights. By passing, pushing, pulling, and turning load carriers, it delivers materials to assembly or picking workstations. The modular gripping concepts based on the same robot platform allow, for example, switching between standard boxes, hazardous goods containers, and other load carriers.

The simulation-based artificial intelligence behind evoBOT® serves as a cyber-physical twin: teams first test adaptations of the platform to new warehouse layouts or load cases in the simulation. The solution exemplifies mobile robots in intralogistics; in operation, evoBOT® acts as a collaborative robot alongside employees. This creates a scalable foundation for efficiently developing autonomous mobile robots for various warehouse and production environments.

Researchers at Fraunhofer IML and the Lamarr Institute first train autonomous agents in a simulation environment before transferring them to real robots. In this “school for AI,” they vary sensor values, floor friction coefficients, structures, and load scenarios to cover a broad spectrum of logistical situations. Reinforcement learning is used to optimize behavior in the logistics space.

In addition to individual robots, the teams are investigating robot swarms. Research into robot orchestration aims to increase flexibility and reduce disruptions. In parallel, they are working with the evoBOT® and O3dyn platforms on digital twins that bridge the sim-to-real gap and enable nearly identical execution of movement commands in simulation and reality.

evoBOT® brings together several areas of expertise at Fraunhofer IML. The platform combines robotics, simulation-based development supported by AI, and digital twins into a seamless development and testing environment. Researchers use state of the art GPUs and motion capture to simulate highly complex motion sequences in real time. They calibrate the simulation using measurement data from the real vehicle. This gradually creates a cyber-physical twin that decouples hardware and software development and shortens development times.

The deep reinforcement learning method, as tested at the Lamarr Institute for Machine Learning and Artificial Intelligence, prepares evoBOT® for a wide range of logistical situations in simulations. Variable physical properties, load scenarios, and environmental structures train robust behavioral strategies. The Open Logistics Foundation aims to drive digitalization in logistics and supply chain management at companies using open-source applications.

Sponsors of the “Silicon Economy” research initiative and
the “Digital Testbed Air Cargo (DTAC)” project