Perception Unleashed: Achieving Safety, Efficiency, and Awareness in Human-Robot Collaboration 

In recent years, the advancement of human-robot collaboration (HRC) has deeply changed the paradigm of industrial scenarios. Indeed, new industrial settings are characterized by collaborative cells, where human and robots work in close contact without the aid of physical barriers. However, to unlock the full potential of HRC, it is crucial to enrich the robot capabilities giving the ability to sense, interpret, and respond to the surrounding environment. 

In this context, perception systems play a pivotal role. Indeed, by merging and exploiting the data coming from multiple sensors it would be possible to control the robot in a more efficient way, adapting to the context or to the human operator that is collaborating with.

When two human operators collaborate, they typically comprehend each others and work together to overcome individual challenges. When performing a load transportation, for example, they assume a posture that makes the lifting of the weight more easy, i.e. trying to improve their ergonomics. Or if one of the two operators is tired due to a high cognitive workload, the other operator can compensate helping it. Thus, it is crucial to endow the collaborative robot with the capability to detect these human-centered factors and to develop control strategy that are able to adapt to such scenarios. 

Furthermore, to emulate the collaboration within human operators, the robotic system must enable also the possibility of a natural, bilateral exchange of forces that convey kinesthetic and tactile information. This phenomenon, known as physical Human-Robot Interaction (pHRI), is instrumental. Besides this classical channel, the exchange of cutaneous tactile data could be expanded using appropriate devices. This further facilitates the transmission of information sensed or processed by the robotic system, enhancing the achievement of collaborative tasks. The coordinated motion is thus influenced by human tactile and kinesthetic sensory systems and their dynamic interplay. It is then possible to leverage this data to allow the robot to collaborate in an effective way.

Such efficient and flexible control strategies, however, do not guarantee that the overall robot behavior is safe. Safety is regulated by ISO/TS 15066, which defines four different collaborative modes. The most important are speed and separation monitoring (SSM), which does not allow a contact, and power and force limiting (PFL), which regulates the possibility of contacts. To ensure compliance with SSM limit, the overall robotic system must be endowed with a system that is able to detect the human operator in the scene. This system could be ideally leveraged to improve the performances, exploiting dynamic planning strategies. When considering the PFL, instead, it is important to develop control strategies that are able to detect when a collision with the human operator occurs. Once the collision is detected, it is possible to activate control strategies that ensure reactive behaviours, ensuring safety.

During this workshop, our objective is to explore these fundamental aspects by inviting five esteemed professors, each renowned in their respective fields.