Kognitive Systeme

Person-Recognizing Autonomous Transport System (PATSY)

A large part of hospital staff's working time is spent on simple transportation tasks. Therefore we are developing an autonomous transportation system with person recognition, in cooperation with the company E&K Automations GmbH. With this system we want to relieve hospital staff of their transportation work. Most current systems are unable to operate in environments shared with visitors and patients. Through modern 3D sensors, intelligent person detection and automatic path planning PATSY's goal is to fill this gap.

The automated guided vehicles (AGVs) are supposed to move around in hospitals without the use of additional markers or guiding tracks. They should also cope with unexpected obstacles like people, boxes, beds, etc. and autonomously find alternative paths. This will enable the system to autonomously transport containers all the way from the basement into the ward.

The robots used for this project are developed by our project partner E&K Automations GmbH. They are built with a low height, which allows them to drive below containers. Through a strong mechanic they are able to lift a whole container with a maximum weight of 250 kg. Every robot is equipped with 2D laser scanners and 3D sensors to sense the environment.

Fig. 1: Automated Guided Vehicle developed by E&K Automations GmbH

Current Focus

Localization and Obstacle Detection

Currently we are focusing on robust localization and reliable obstacle detection. Sensor data for both tasks and its interpretation is visualized in figure 2.

Data from the 2D laser scanners is analyzed to determine the position of the robot. We use state-of-the-art localization algorithms and adapt them to make them work robustly in a hospital environment. Therefore we need to detect common dynamic objects like legs and remove them from the scan used for localization.

Data from the 3D sensors is used to detect obstructions like tabletops and shelfs, that are invisible to the other sensors

Fig. 2: A 2D-Laser scan (red) used for localization. Segmented point cloud of obstructions (gold) and floor (anthracite).

Project Partner:

This work is funded by the Federal Ministry of Education and Research int the context of the founding program "ICT 2020 - Research for Innovation" in the field of service robotics.

References

[1] Sebastian Buck and Andreas Zell. Cs::apex: A framework for algorithm prototyping and experimentation with robotic systems. Journal of Intelligent & Robotic Systems, Apr 2018. [ DOI | details | link ]
[2] Sebastian Buck, Richard Hanten, Karsten Bohlmann, and Andreas Zell. Multi-sensor payload detection and acquisition for truck-trailer agvs. In Robotics and Automation (ICRA), 2017 IEEE International Conference on, Singapore, 2017. [ details ]
[3] Sebastian Buck, Richard Hanten, Karsten Bohlmann, and Andreas Zell. Generic 3d obstacle detection for agvs using time-of-flight cameras. In Intelligent Robots and Systems (IROS), 2015 IEEE/RSJ International Conference on, pages 4119 -- 4124, Daejeon, Korea, October 2016. [ DOI | details ]
[4] Richard Hanten, Sebastian Buck, Sebastian Otte, and Andreas Zell. Vector-amcl: Vector based adaptive monte carlo localization for indoor maps. In Intelligent Autonomous Systems (IAS), The 14th International Conference on, Shanghai, CN, July 2016. [ details ]