Operating from the European capital for robotics, automotive industry and ICT, we at the Technical University of Munich (TUM) are contributing to the growth of leading companies in our ecosystem. Our competence in identifying technology gaps, defining technological roadmaps and offering innovative, cutting edge solutions place us at the head of the industry 4.0 revolution.
With over 30 years of experience in sensor based cognitive robotics and a strong network of industry partners, we can serve as the enablers for constructing a unique platform for high tech robotics development in Singapore. Such a platform will help attract and harness local talent and enable Singapore to add technology development for advanced robotics and systems to its list of expertise. In addition, through our extensive track record in establishing and executing large scale European projects, we have demonstrated our ability to attract large conglomerates to invest knowledge and capital in technology development and its ultimate transfer to society through incubation and enterprise.
Efficient and reliable automation of aviation cargo handling is not trivial, thereby justifying the formulation of the Aviation Challenge by CAAS. The high-mix high-volume nature of the problem exposes several complex issues which state of the art advanced robotics cannot easily overcome. The build-up and break-down of aviation cargo pallets involves handling a very wide range of boxes in terms of size, shape, material and mass. Therefore, traditional sensing techniques and robotic palletizers are incapable of addressing this high-mix problem. Furthermore, a one size fits all solution is impractical and therefore opens the doors for pushing the boundaries of robotics in terms of sensing, planning and control.
Currently there is no existing off the shelf system which can automatically palletise High-Mix High-Volume cargo. Accurate sensing of incoming cargo for reconstructing its geometric, dynamic and material properties is quite a complex challenge. Conventional computer vision techniques do not scale in term of range, robustness and speed. Therefore, developing a compact, scalable and robust sensing system for aviation cargo is essential.
Furthermore, reliable manipulation of high-mix cargo poses several challenges on the design and control of the handling system. These are twofold: robust gripping system and precision motion control with optimized dynamics. Given that a one size fits all gripping system is not pragmatic, identifying the optimal (largest) subset of the problem space which can be managed by a minimal set of custom designed grippers is an added challenge. Precision motion control with optimal dynamic is a must for ensuring the integrity of the handled cargo during manipulation. In addition to sensing and control, high level planning and optimization of the build-up process while assuring safety and variable levels of autonomy and intuitive interfaces for humans in the loop operation are also challenges.
The solution we are developing as a part of the Aviation Challenge will enable systematic, flexible and scalable automation of the aviation cargo handling process. It will be realized using the vital ingredients which define intelligent systems: perception, cognition and action.
The system uses advanced 3D sensing technology and multi-modal fusion to perceive the environment. It generates knowledge about the state of the environment in real-time by spatio-temporal localisation and classification of the dynamic process entities.
The process execution is facilitated by a cognitive architecture which uses input from the perception system and advanced computational models to generate an optimized sequence of actions. The actions are mapped to the available actuators in the system such that the process is optimized in terms of time, energy and cost.
The decisions made in the cognition loop are grounded to physical actions through the robot system. Furthermore, actions involving interaction with humans are achieved through multi-modal user interfaces.