Periodic Reporting for period 1 - TOAST (Touch-enabled Tactile Internet Training Network and Open Source Testbed)
Période du rapport: 2023-03-01 au 2025-02-28
The Horizon Europe MSCA Doctoral Network project, TOAST, focuses on the emerging, interdisciplinary field of the Tactile Internet. It will train 10 doctoral candidates (DCs) through a comprehensive programme combining scientific, technological, and transferable skills training via lectures and hands-on exercises. International and inter-sectoral research mobility across six European countries will expose DCs to diverse environments and cutting-edge research. By bringing together leading academic and industrial partners with complementary expertise, the project aims to develop holistic solutions for the Tactile Internet. Aligned with EU strategic goals in digitalisation, sustainability, and next-generation connectivity, the project also contributes to the Green Transition through innovative digital technologies.
The research objectives of TOAST include (1) Develop a multi-purpose Open Source TI Testbed based on the IEEE P1918.1; (2) Design haptic technologies for communication including new kinaesthetic and tactile codecs. (3) Joint optimisation of compression, control and networking. (4) Design Edge Intelligence Engine to enhance reliability and agility in TI services. To achieve these objectives, the research program is structured into four dedicated technical work packages: WP1- Use cases requirements and testbed, WP2- Haptics for communication, WP3- Agile communication and networking, and WP4- TI Edge Intelligence Engine.
• WP1 – Use Case Requirements and Testbed: Defines the requirements for selected use cases and coordinates implementation, prototyping, and demonstrations.
In WP1, three primary use cases were identified as most relevant to the TOAST project: (1) bilateral teleoperation with haptic feedback, (2) immersive virtual reality, and (3) skill transfer and remote learning through demonstration. Key technical requirements including stability, maximum force feedback, stiffness, response speed, impedance range, manipulability, and synchrony between video and haptic modalities, were summarised based on the partners’ expertise and an in-depth review of the literature.
• WP2 – Haptics for Communication: Develops advanced kinaesthetic and tactile codecs, and energy-efficient wearable devices for remote touch.
In WP2, a novel kinaesthetic codec has been proposed, aligned with the IEEE 1918.1.1-2024 standard, with an aim of ensuring stable and high-fidelity force and motion feedback under varying network conditions. It introduces key innovations that improve haptic communication efficiency. A kinaesthetic testbed was created for controlled evaluation of the codec’s performance under realistic network scenarios. This forms a robust methodology for kinaesthetic data compression and adaptive control. Additionally, a new multimodal tactile codec has been developed to compress tactile signals from multi-point contacts for all five tactile dimensions. By integrating a comprehensive psychohaptic model, it preserves perceptual quality and enables rich tactile signal transmission beyond surface roughness. Furthermore, a low-power, wearable, user-friendly hardware device for remote touch is under development. Designed for seamless integration into virtual reality systems, it supports intuitive visuo-haptic immersive experiences through haptic codec integration.
• WP3 – Agile Communication and Networking: Designs agile communication solutions for Tactile Internet over 5G/B5G networks.
In WP3, extensive measurements have been conducted across diverse network configurations for a set of representative TI services using both a cloud-native proof-of-concept testbed and a fully operational 5G standalone private network. A thorough evaluation of open-source tools and frameworks has supported the development of an agile communication testbed, incorporating both 5G core and radio access network components in alignment with industry standards. Furthermore, an end-to-end network slicing framework has been designed to meet the stringent and heterogeneous requirements of multimodal Tactile Internet services.
• WP4 – TI Edge Intelligence Engine: Leverages edge computing and AI to deliver intelligent support for Tactile Internet services.
In WP4, a suite of advanced Edge Intelligence functionalities has been developed or is currently under development as part of the TI Edge Intelligence Engine, aimed at addressing key technical challenges in the Tactile Internet. These include more efficient radio resource allocation, mitigation of packet delay and loss, enhanced remote scene reconstruction, and seamless human–machine collaboration. For instance, a machine learning-based traffic prediction is proposed for efficient resource allocation; a deep reinforcement learning-based network slicing is proposed haptic-video traffic, delivering significant gains in user experience; a novel lightweight, memory-free continual learning method is proposed for cross-task learning; and a touch-augmented 3D Gaussian Splatting framework is proposed to improve remote scene reconstruction in visually degraded environments.
These research outcomes play a pivotal role in enabling the key use cases in TOAST project. Additionally, datasets, such as a kinaesthetic dataset, have been generated and made publicly available. Together with the developed testbeds, these open-source resources provide valuable assets for advancing future academic research and facilitating industry adoption.