The Rossini Platform
The developed components are expected to be integrated in to the ROSSINI Platform architecture. The Platform can be represented as an integrated set of layers, each related to a specific dimension/function:
The Sensing Layer will combine information from safe and non-safe sensors in a fusion module to feed the Safety Aware Control Architecture.
The Perception Layer, through the employment of artificial intelligence techniques, will generate a Semantic Scene Map integrating geometric and semantic information, which will in turn create a set of virtual “Dynamic Shells” for safety, surrounding each object in the scene.
The Cognitive Layer will be provided by a high-level scheduler, capable of dynamically planning a set of cooperative actions that the robot needs to execute, and to update them when the working environment conditions, captured by the Semantic Scene Map, change.
The Control Layer will interpret the high-level action to execute and will generate the most efficient and safety preserving low-level plan for the robot, thus optimizing trade-off between safety and productivity in the workcell.
The Actuation Layer will encompass a novel concept of manipulators with built-in safety features, capable of reducing the separation distance between the man and the operator when performing collaborative applications, thus increasing the degree of freedom for robotic applications design.
The Human Layer will ensure the inclusion of human-related factors from the early design phases of collaborative applications design, and the constant monitoring of factors influencing job quality during robotic operations.
The Integration Layer will provide integrators with a set of tools and guidelines to ensure inherent safety in design of HRC applications, and to speed up application configuration and reconfiguration.
Moreover, the platform will include also a set of methodologies and guidelines to improve application design and risk assessment in HRC. Recent research studies (R. Behrens, N. Elkmann, and H.-J. Ottersbach 2012) show that the difference between free and clamping impacts depends on how the involved robot and human masses are distributed. ISO/TS 15066 already provides a scaling factor that allows for switching measured impact results between both contact cases. The factor only applies if the effective masses, colliding at the contact point, are given. A method to estimate the robot mass is also available in ISO/TS 15066. It takes all link masses and their configuration account and estimate the mass at the considered contact point (in most cases the robot TCP). Besides this estimate, each robot manufacturer offers models with higher precision. Instead of providing a similar method for estimating the effective mass of the human body, ISO/TS 15066 recommends using the single weights of the particular body parts and neglects the body kinematics. From a scientific standpoint, this approach will lead to wrong and highly biased estimates that have the potential to a wrong risk evaluation and is therefore not accepted by official bodies. To ensure reliable estimates of body part masses, the ROSSINI Platform will include a simplified human body model that replicates the kinematics and mass distribution of a 50th percentile human. The development work to be carried out will include the development of the kinematics of the model, the integration with other available studies and the conversion the model in an algorithm. The inertia parameters will be derived from the result of a collision study with volunteers which was carried out in 2012 (ethical approved – see R. Behrens, N. Elkmann, and H.-J. Ottersbach 2012). The study goal was to determine the difference between constrained and unconstrained impacts. The results of this study can be considered as valid and enables the model to create reliable results.
ROSSINI Demonstration Activities
The ROSSINI Platform will be demonstrated into 3 industrial environments up to TRL6. The use cases have been chosen trying to have the widest possible span in terms of application sector, tasks to be executed technologies to be deployed.
Use Case #1 – Domestic Appliances Assembly (WHIRLPOOL)
In the WHIRLPOOL use case, ROSSINI will deploy the adaptive features of the Human Layer and the Safety Aware Control Architecture in a highly challenging context (continuous flow line) in terms of working speed. The manipulator will be chosen among commercial products, to demonstrate the platform potential to interact with and be wrapped around third party technologies.
Use Case #2 – Electronic Components Production (SCHINDLER)
In the SCHINDLER use case, ROSSINI will demonstrate the features of the Collaborative by Birth Robotic Manipulator, delivering a low payload robotic arm for electronic components production, integrated with the other platform components (sensing system, controller, etc.). Given the user’s high production variability, the promised ROSSINI performance in terms of production reconfiguration cost savings will be tested.
Use Case #3 – Food Products Packaging (IMA)
In the IMA use case, ROSSINI will deploy a medium payload manipulator mounted on an AGV (mobile robotic platform). The Human Layer and the Control Architecture will be confronted with the challenge to manage navigation operations as well as other operations. Being IMA also a robot integrator, the demonstrator will be easily replicated.