However, despite these developments and improvements, hard mate- rial robotic machining with conventional robots and control approaches remains so far a challenging problem, especially for small-batch produc- tion of complex parts. During hard-material cutting, the position devi- ations and tool chattering effects result in a not satisfactory quality of the milling surfaces. The finishing of surfaces by using grinding tools to improve the manufacturing quality requires even higher positioning accuracy, which is beyond robot position servo limits.

From academic research, current efforts mainly handle the men- tioned practical problems, addressing different integrated approaches based on robot signatures, simulation of the robotic systems and cutting process, various advanced robot control algorithms, as well as high-tech solutions based robot-dynamic motion tracking and real-time correc- tion of the robot poses and high-dynamic compensation on the tools [4–10,12,13].

The integrated machining cell design in conjunction with modular reconfigurable design, robot specific planning that considers real motion limitations and performance, efficient cell alignment methods have been proven in [5] as an competent framework to improve feasibility of the robot machining process.

Compensation for robot positioning errors due to dominant elastic deformation in joints (transmission mechanisms) represents a challeng- ing issue. An off-line model-based approach to predict, analyze and com- pensate for the elastic robot deformation under cutting process forces in milling applications has been presented in [6]. The real-time control compensation of quasi-static elastic deformations in robot transmissions using force sensing and deviations computations, based on joint elastic- ity models and Cartesian deformation mapping, has been recently pro- posed in [7,8]. In the initial experiments the improvement of the milling path roughness accuracy up to 50% was demonstrated.

The vibration/chattering effects especially during the interaction be- tween robot-carried tools and material represent one of the major hur- dles preventing the robotic-machining process from being applied in hard materials [11]. Vibration analysis based on the elasto-dynamic robot model in Cartesian space has demonstrated that the chattering ef- fects depend on robot configuration (due to variable apparent Cartesian robot stiffness and inertia). The low frequency mode coupling chatter can shake the entire robotic structure and cause damages to the system. By selecting proper machining direction and relative tool motion these effects may be reduced and suppressed.

The usage of force-sensing in real-time control has been identified as a very promising approach for both programming and real-time control. A force-based approach for lead-through programming and controlled material removal rate has been presented in [12]. The improved robot performance by applying dynamic-force control in the robotic grinding, deburring and drilling has recently demonstrated in [6]. However, de- spite some robot producers (e.g. ABB, KUKA) offer add-on force-control functionalities, the high dynamic and force control have not been pos- sible to be implemented in the industry in a high performance flexible manner due to quite conservative marketing approach. The robot pro- ducers commonly prefer to integrate sensors, to tune controller and to make programming for a specific application. Such policy reduces the robot flexibility and re-programmability by the end-users. The program- ming of robots in term of interaction force is difficult and not well un- derstood for the users, from the view of different applications.

The lack of understanding of interaction tasks programming, espe- cially complex machining, as well as the missing of a widespread inter- action control design approach for industrial applications are recognized [13] as the main bottlenecks for a wider application of force control in the industry.

The specific robot application problems related to the small batch production have been investigated in  [1,2].