Login Form

IP 287513
call FP7-ICT-2011-7
SAPHARI is funded under the European Community's 7th Framework Programme


Short videO overview

Punching holes with robot HASy

Playing drums with qbmove maker pro


Use of sensitive skin on robot manipulators


Safe physical human-robot collaboration


Close and wide range monitoring


Multi-modal interaction and kinaesthetic teaching 


Safety features in the hospital use case


Human-robot interaction in an industrial use case

More videos can be found in our Video Gallery.


Human safety. Only few years ago, international standards required robots to be separated from humans and to stop as soon as an operator entered their workspace. Research on human safety in SAPHARI has paved the way to understanding injury mechanisms and devising countermeasures in hardware and software, which will possibly be reflected in a near future in relevant international standards for pHRI and into products. 
Deliverable D1.1.1 - Human injuries in robotics 
Deliverable D1.3.1 - Safety monitoring framework and safe control strategies
Deliverable D1.4.1 - Existing standards and recommendations for future safety standards in robotics

The mechatronic development of assistant robots is a central task for achieving truly human-centred robotics. Two main design paradigms have been followed in SAPHARI. The first is based on the technology of the DLR torque-controlled robots, which has now become a commercial product by KUKA. The second design philosophy exploits the new solution of Variable Impedance Actuators (VIA), leading to effective multi-dof robot prototypes.
Deliverable D2.1.1 - New actuators for the robot companion
Deliverable D2.2.1 -­ Low-­cost modular VSA platform
Deliverable D2.3.1 - Compliant robot body
Deliverable D2.4.1 - Prototypes of safe and interactive tools and hardware safety concept of workspace 

Perception and control. SAPHARI has brought to maturity methods for detecting and tracking humans that move in the robot vicinity. Moreover, we developed a range of control strategies and low-level reactions (reflexes) that enable a robot assistant to react properly, quickly, and safely to intended or unintended physical interaction. A major effort has been also devoted to the feedback and energy-optimal control of VIA-based robots.
Deliverable D3.1.1 - Combined collision avoidance, detection, and reaction
Deliverable D3.2.1 - Control and estimation algorithms for VIA robots
Deliverable D3.2.2 - Experimental validation of control laws for multi-dof VIA manipulators
Deliverable D3.3.1 - Safe dynamic control laws for redundant robots
Deliverable D3.4.1 - Intentional human-interaction control of compliant robots
Deliverable D4.1.1 - Collision detection methods by means of sensor fusion
Deliverable D4.2.1 - Prediction based algorithms for wide range monitoring
Deliverable D4.3.1 - Tracking of human motions and object interactions
Deliverable D4.4.1 - Detection and isolation of sensors/actuators faults

Learning and interpretation of human actvites by the robot. A robot companion must be able to adequately interpret and dynamically react to human actions. In SAPHARI, learning and semantic interpretation on of human behaviours have been cornerstones for making the robot aware of human actions at a cognitive level, and to optimally react to human intention.
Deliverable D5.1.1 - Gesture and grammar descriptors of human motion and statistical gesture parser
Deliverable D5.2.1 - Intuitive motion learning and iterative motion refinement
Deliverable D5.3.1 - Learning of force patterns and impedance behaviors 

Task reasoning and situational reaction. The planning approach developed in SAPHARI uses perceptual information and scene interpretation for adapting on-line the robot plans, so as to complete the task by taking into account the human actions. Depending on the task and on the human intention, the planner switches between complete autonomy in the absence of humans to different levels of collaboration modes. 
Deliverable D6.1.1 - Reactively adaptable motion plans for real-time collision avoidance
Deliverable D6.2.1 - Multimodal reactive motion generation
Deliverable D6.3.1 - Action generation patterns and search algorithms
Deliverable D6.4.1 - Interface framework for multimodal safe interaction
Deliverable D7.1.1 - Final models and algorithms for collaborative activities
Deliverable D7.2.1 - An implemented human-aware navigation and placement planner
Deliverable D7.3.1 - An implemented human-aware manipulation planner
Deliverable D7.4.1 - Planning of smooth motion-force transition tasks
Deliverable D7.5.1 - Specification of a human-aware robot controller
Deliverable D7.5.2 - Prototype of the human-aware robot executive controller 

The project focused on two industrial use cases that explicitly contain deliberate physical interaction between a human and a robot co-worker, driven by the industrial partners AIRBUS and KUKA respectively. A professional service scenarios in hospitals, in which medical staff and an assisting robot interact closely during daily work, was developed under the supervision of DLR

Demonstration of the AIRBUS use case at the final review meeting in Augsburg 

Demonstration of the KUKA use case at the final review meeting in Augsburg

Demonstration of the hospital use case at the final review meeting in Oberpfaffenhofen


Total number of publications distributed per type  Publications distributed per type and year

The complete list of publications can be found here.

Institute of Robotics and Mechatronics (DLR)