A Smarter Way for the Visually Impaired to Navigate with PathSense
For many visually impaired individuals, navigating the world independently can be incredibly challenging. White canes, while essential, have limitations, especially when it comes to detecting obstacles above waist level. PathSense aims to bridge this gap by integrating AI-driven object detection into a compact crossbody device. Using cameras and sensors, the device continuously scans the user’s path, processes the data through cloud-based AI, and provides real-time audio feedback to prevent potential hazards.
The team behind PathSense is a group of young CMKL sophomores who are passionate about making assistive technology more affordable, effective, and sustainable.
“One of our goals for PathSense is to ensure that our technology is affordable since everyone should have the right to improve their quality of life” the team explains.
This commitment has guided their development process, influencing their choice of hardware, AI architecture, and business model.
PathSense operates on a cloud-based AI model that analyzes images captured every two seconds by a wearable camera. This approach allows the team to keep hardware costs low while improving computational efficiency. Instead of relying on expensive onboard processing, the AI model runs on a cloud server, delivering audio cues to users in real-time.
To improve accessibility, the team also focused on enhancing their text-to-speech (TTS) system. Initially, the device relied on built-in mobile TTS features, but with a shift towards direct cellular connectivity, they needed a new approach. After evaluating various solutions, they implemented a cloud-based caching system that progressively reduces latency by storing frequently used words and phrases.
Additionally, the firmware and connectivity pipeline saw significant upgrades. The team moved away from traditional PPP-based SIM connectivity, which required tedious manual configurations, and instead adopted ModemManager, a more streamlined approach that simplifies network switching which is essential for future scalability.
Another crucial step was data collection. The team trained their AI using a mix of real-world and synthetically altered images to simulate different environmental conditions like low light and varied angles. “We also augmented the training data by lowering brightness to mimic darkness,” they shared. While current models perform well in controlled environments, future improvements aim to enhance performance in challenging conditions like rain.
One of the most surprising lessons for the team was realizing the long-term potential of PathSense.
“Since this project started as a Year 1 Research and Development, up until the AIS Jump Hackathon in 2024, we never would have realized the potential that our product had,” they reflect.
The hackathon forced them to make a decision—was this just a university project, or could it grow into a full-fledged company? They chose the latter, and now their roadmap stretches years ahead.
Another critical realization came from the challenge of securing funding while balancing business sustainability. The team has already secured 25,000 baht from CMKL University and is pitching for a 200,000 baht government grant from DEPA. However, company registration issues have temporarily delayed access to additional funding, forcing them to rethink their financial strategy.
Despite these hurdles, they remain focused on their next big milestone: user testing with the Thailand Association of the Blind Foundation. “Our first major milestone with funding would be to carry out user testing so that we can keep reviewing the customers’ expectations,” they emphasize.
PathSense stands out by balancing affordability, usability, and innovation. Unlike other high-tech mobility aids, which are often prohibitively expensive, PathSense prioritizes cost-effective components without sacrificing functionality. Compared to traditional white canes, it offers a significant advantage by detecting above-waist obstacles, preventing accidents that current aids fail to address.
Additionally, the compact design allows for seamless integration into daily life, an aspect the team believes is crucial.
“We pride our product on how it easily integrates with the users’ daily lives due to its small form factor,” they explain.
As PathSense moves forward, its biggest challenge remains refining the AI model to adapt to diverse real-world conditions while ensuring that the product stays accessible to those who need it most. If the team secures full funding, the next six months will be dedicated to expanding user testing, refining the prototype, and preparing for a wider launch.
With a clear vision and unwavering determination, PathSense is on track to redefine independent navigation for the visually impaired—one step at a time.
