A robot physically links a skilled teacher and a novice learner, and the strength of that connection determines how quickly the learner picks up a new motor skill. This project addresses a gap in understanding how to design physical human-robot interactions for learning. Previous work showed that an *asymmetric* connection—where the robot transmits different levels of force or guidance to each person—improves performance in a shared task. But no one has systematically tested which type of asymmetry best accelerates skill acquisition over time, or whether those lab findings transfer to a real classroom setting. If the research succeeds, it could change how children learn handwriting. Instead of a teacher simply demonstrating letter shapes, a robot could physically guide the child’s hand while simultaneously giving the teacher haptic feedback on the child’s movements. The same principle could apply to other fine-motor skills—playing a musical instrument, using surgical tools, or operating industrial machinery—where a trainer’s physical intuition is hard to verbalise but easy to transmit through touch. The project is applied, not fundamental science. It takes a known laboratory effect and tests it in a practical context with a clear end user: primary school children and their teachers.
View original technical description
Considering the notable performance benefits associated with the asymmetric connection (Noccaro et al. 2024) and recognising the importance of optimally designing the coupling connection in learning contexts (Teranishi et al. 2018), the main objectives of this project are: i) to investigate how to take full advantage of the asymmetric connection mediated by robots for the enhancement of human motor learning, studying how learning is affected by the levels of connection and how these can be optimised to improve the acquisition of new motor skills; ii) to apply the knowledge gathered from the use of the connecting robots to a real-life application, specifically to foster handwriting learning in primary school children while interacting with their teacher or caregiver. The outcomes of this project, by revealing the most advantageous approach to leverage haptic support during the learning process, could exert a substantial impact on both the design of novel haptic devices and the improvement of traditional learning procedures. Project implementation The project is structured in two main phases: 1) assessment of the best coupling condition in a laboratory task; 2) assessment of the defined coupling condition in a real-life application, i.e. handwriting learning in children. 1. Assessment of the best coupling condition in a laboratory task. In this phase the experimental setup used by Noccaro et al. 2024, involving robots to connect two people executing a common 3 DoF task, will be used in a set of experiments involving two individuals with different motor skills (modulated by introducing a specific visual noise). The expert (trainer) and non-expert (trainee) users will be coupled by different symmetric and asymmetric connections during a typical learning protocol, comprising multiple experimental sessions and a longitudinal within-subjects design to assess individual learning over time. By evaluating different sets of symmetrical and asymmetrical connections, this phase aims to define the optimal connection for enhancing the learning process to be used in the second part of the project. 2. Transition to a real-life application - handwriting learning. The second part aims to extend the insights gathered from the lab experiments to a real-life application. Specifically, the project will focus on handwriting learning in children for twofold reasons: the high relevance of handwriting in children's learning process (Feder et al. 2007) and in general in human long-life learning, and the high complexity of this task, requiring the acquisition of skills in visual-motor integration, motor control, cognitive and linguistic fields (Feder et al. 2007).
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know