Access Official Technical PDF

Share Summary:LinkedInEmail

Environmental sensor networks are typically designed to collect measurements for analysis, monitoring and decision-making. A research project at Tidmarsh, a wetland restoration site, explored another possibility: translating distributed environmental data into sensory experiences that help people notice processes beyond the reach of ordinary perception. Its central lesson for automation and sensing professionals is that the value of a network depends not only on what it measures, but also on how people can interpret and experience its data.

Environmental sensor networks as part of a living landscape

Tidmarsh was formerly an industrial cranberry farm and became a wetland restoration site. A continuously operating sensing system was deployed there to document ecological conditions before and during restoration, with data made available to researchers, restoration engineers and land managers. The network was also conceived as a platform for public-facing applications, connecting environmental monitoring with new ways of engaging with the landscape.

This broader approach was called a Networked Sensory Landscape: a physical environment permeated by sensing infrastructure, where measurements can support scientific inquiry and shape how people experience a place. The underlying motivation is practical. Many changes in an ecosystem happen too slowly, too quickly, or in locations people are not observing. Instrumentation can capture those changes, but conventional dashboards and data displays are only one way to make them meaningful.

The system followed an end-to-end architecture with three main layers:

  • Collection: distributed sensor nodes, microphones and other sensing sources gathered environmental measurements and sound.
  • Back end: data and audio services made streams available to applications through network interfaces.
  • User interfaces: software and installations translated those streams into experiences for researchers and visitors.

Designing these layers as parts of one system allowed the same infrastructure to support both monitoring and experimental interfaces. The data network was not treated solely as a backend for reports; it became a foundation for exploring how environmental information could be perceived in context.

From sensor data to sensory experience

One challenge in any sensing application is the distance between a measurement and its meaning. A displayed value may be precise, yet still require a user to stop, inspect a screen and interpret what it represents. The Tidmarsh projects investigated sound and vibration as alternative ways to present information, aiming to make selected data perceptible without turning every interaction into a screen-based task.

Earlier experiments included sensor-driven virtual environments, sound installations and vibrating objects. These projects explored different relationships between a person, a physical site and data from that site. In virtual environments, sensor readings influenced how a landscape was rendered. Other installations combined live audio with curated sounds or used vibration to connect stories and environmental information with physical objects.

The projects were not simply about adding effects to data. They examined how people build an understanding of information when it is presented through a sensory medium. A sound can be situated in space, overlap with ambient audio and invite attention without requiring a user to navigate a conventional interface. But the experience also depends on learning: listeners need to develop an intuition for what they are hearing and how it relates to the environment.

Spatial audio and the HearThere prototype

The culminating system, HearThere, used non-occluding bone-conduction headphones to blend a digitally rendered soundscape with the wearer’s natural hearing. It drew on live audio from distributed microphones, along with sonified sensor data. Because the device did not cover the ears, users could continue to hear the physical environment while listening to audio sources associated with other parts of the site.

HearThere also adjusted how virtual sources were mixed and positioned. Its software combined an inferred listening state with output from an AI-based sound classification system. The classification engine could help emphasize identified wildlife sounds or reduce the prominence of wind noise, while the attention model used available sensor inputs to influence audio channels. The result was a changing spatial sound field rather than a fixed, uniform playback mix.

Research into spatial hearing and auditory attention informed the design. Field evaluation at Tidmarsh examined how participants made sense of the experience. A key finding was that perceptual ambiguity was part of the interaction: users sometimes had to distinguish between a sound occurring around them and a sound delivered through the device. With exploration and experience, participants could begin to interpret these blended sources. The design goal was not to eliminate every ambiguity, but to create a setting in which users could develop new ways of listening.

Design implications for sensing and automation

Although the work focused on a restored wetland rather than a production facility, it raises useful questions for engineers and technical decision-makers building sensor-driven systems. Data collection is only one component of an effective sensing architecture. The choice of interface affects which patterns users notice, how quickly they can interpret conditions, and whether information supports exploration as well as specific tasks.

Several principles emerge from the project:

  • Design the full data path. Sensor placement, communications, backend services and presentation should be considered together. An interface cannot compensate for missing, inaccessible or poorly structured data.
  • Match the medium to the context. Sound and vibration can communicate information without relying exclusively on visual displays, but they need to fit the environment and user’s activity.
  • Make data interpretable over time. Users may need repeated exposure to associate a sensory pattern with a measurement or event. The system should support learning rather than assume immediate comprehension.
  • Use automated classification thoughtfully. Machine-generated labels can shape what users hear, so classification output and attention controls become part of the interface logic—not merely backend processing.
  • Preserve the relationship to the physical setting. In a place-based system, digital information is most useful when its presentation helps users understand where it comes from and how it relates to their surroundings.

Conclusion: Build sensing systems people can make sense of

The Tidmarsh work illustrates an expanded role for environmental sensor networks: they can support ongoing measurement while also enabling new forms of situated perception. By combining distributed sensing, live audio, classification and spatial presentation, HearThere explored how data from across a site could be experienced alongside natural hearing. The approach does not replace conventional monitoring or analysis; it adds another way to engage with information.

For industrial automation teams, the practical implication is to treat sensing, data services and human-machine interfaces as an integrated design problem. A useful system must collect reliable information, deliver it to the right application and present it in a form people can interpret in context. Whether the setting is a wetland or an industrial operation, thoughtfully designed interfaces can help turn streams of measurements into observations that inform understanding and action.

Access Official Technical PDF

About The Author