Richard Doell

Developed a technique for symbol grounding across multiple modes, mainly audio and visual, in the form of speech and images while working on my graduate degree. The technique was created using unsupervised methods to find cluster centers between these two domains.

This bidirectional process recognized images and played and understood the associated and accumulated speech. This led to recovering the associated image, emulating how humans identify and correlate objects between different sensory modes.

Helped coordinate and led the development of an open-source, augmented reality robotics platform to enable lower-cost experimentation and learning. At the time, getting into robotics was expensive, mainly due to the price of sensors.

I worked with a local Python developers' group, Makerspace, and the Space Foundation Discovery Center. We built a simulation and augmented reality room, a target platform, and software to run robots and demonstrated the project at the Colorado Springs Mini Maker Fair.

Designed and developed a Rust-based semantic network tool, a system for visualizing and processing the semantic database. Inspired by older semantic web concepts of labeled, understandable links, the tool is aimed more at data curation than application support.

As a tool that's part-database, it enables users to capture information about any domain and, with customizable links, network that information to the rest of their database. This mimics the way that memory works, especially recalling information that has a meaningful association. The benefit of the format is that it enables data exploration via techniques like Zettelkasten, as well as both statistical and traditional graph processing. This allows users to find new connections between elements of their web, leading to new ideas and inventions.