Interconnection

Currently, interactions are typically centered around powerful and independent computing devices. A pervasive interface requires interactive everyday things, which will introduce unacceptable deployment and maintenance efforts. Interconnection techniques reduce such efforts by enabling users to redistribute energy, computation, and I/O resources among computers and everyday things. Only necessary functions like sensing and feedback are kept on the resource-constrained things, while demanding functions like computing and power are offloaded to the resource-abundant computers.

Figure 1. TouchPower is an Interaction-based Power Transfer (IPT) technique. It transfers DC power through contact between electrodes on the glove and the object.

Powering off-body objects with on-body energy sources can significantly reduce user maintenance efforts. I invented the concept of Interaction-based Power Transfer (IPT), which leverages the contact and closeness between user and object during interaction to establish power transfer channel. IPT is especially suitable for devices that only need to be powered during interaction (e.g. mouse, remote controller). The concept was validated with a glove-based IPT prototype TouchPower (Figure 1), which transfers DC power through contacts of electrodes on the glove and objects. With careful design of the transfer interface, energy can be distributed without affecting the original interaction.

Figure 2. Tap-to-Pair enables wireless device association with minimal resource requirements.

I also develop human-centered association techniques to initiate interaction from the more available everyday things. Tap-to-Pair supports spontaneous device association based on temporal correlation of two signals. Users can tap on an IoT device to induce periodic wireless signal strength changes, which is then correlated with the blinking patterns of target devices for association (Figure 7). A follow up work proposes a 2D design space and design guidelines for blinking patterns by applying Bayesian models of user tapping behaviors. Such optimization enables the technique to support robust selection among more targets. A functional Tap-to-Pair application for Linux systems is open sourced on GitHub.

Go to Sensing Tags
Go to Finger Wearables

Avatar
Tengxiang ZHANG
Assistant Research Scientist

I use my hardware and software skills to build interfaces between the physical and digital world.