One the objectives of the ph-coding project was the direct application of the concept of predictive coding to the domain of haptic interfaces. Haptic interfaces combine movement and tactile information to create new types of human machine interfaces with applications in many domains such as vehicle cockpits, virtual reality technologies, accessibility, well-being, and many more domains. Partner company Actronika SAS [https://www.actronika.com/] has assiduously applied the results of the ph-coding project to create new types of haptic interfaces that provide movement-induced sensations characterised by a high degree of realism and which do not require any training from their users nor any attentional engagement to be used efficiently.
The study of the brain circuitry mechanisms underlying haptics perception has led to a foundational technology development in Artificial Intelligence based on predictive coding perception, rather than on machine learning-style inference. It is a breakthrough technology that promises qualitative advances in AI, and has the potential to outclass traditional artificial neural network (ANN) technology in many domains. The objective is to provide a platform to achieve Articial General Intelligence defined as including artificial perceptual processes into the inferential process. This breakthrough technology will make AI more useful and amenable to real world applications. IntuiCell AB [https://intuicell.com/] is a spin-off company created from ph-coding results that implements predictive coding technology. Being based on self-discovered, self-acquired knowledge in articial neural networks, this technology creates intelligent systems that are truly autonomous and applicable to a wide range of problem domains.
Both the skin technology and the network technology, as well as the central principle of predictive coding, have been implemented in a haptic intelligent robot system that can learn to optimally interact with its environment during movement.
As a bridge to the domain of robotics control, and the path towards autonomously intelligent robot systems, ph-coding work has also generated an interfacing technology, building on similar network technology as the novel AI perception technology above, but which map information about the domain of the body, or the robotic ‘plant’ configuration, to the domain of dynamic network representations. This generic circuitry model brain sensorimotor control within ph-coding represents a first step towards creating a new technology platform for future better integration between AI network technology and haptically intelligent robotic systems.
