Enabling High-Precision Crowd-Sourced Mapping at Scale
Hivemapper partnered with Hellbender to build a compact, AI-enabled camera platform that brings advanced street-level mapping to the edge. The result? A solution that reduced per-kilometer processing costs by 90%.
Partner Profile
Hivemapper is a San Francisco startup focused on building and delivering high-precision mapping solutions at scale using their proprietary dash-mounted, AI-enabled Bee camera. Their mission is to make real-time street-level data collection faster, more accurate, and cost-effective. To achieve this, Hivemapper partnered with Hellbender to develop an innovative camera system that fundamentally redefines the economics of mapping.
Project Challenges
Hivemapper needed a rugged, real-time mapping platform that could match the accuracy and performance of cloud-based systems—without the high costs. Key challenges included:
- Compact AI Integration: Incorporating sensors and advanced vision processing in a small, durable housing that could withstand a broad range of temperatures and vibration profiles.
- Depth Perception: Ensuring accurate stereo vision for complex street-level imagery.
- High-Precision GPS: Supporting both L1 and L5 signals for mapping accuracy.
- Reliable Connectivity: Enabling real-time data transmission from any location.
- Designing for Scalability: Ensuring that the Bee’s design, integration, and bring up procedures were appropriate for scaled production
Solution & Results
Working closely with the Hivemapper team for feedback, Hellbender engineered the Bee Camera System, a compact, AI-enabled platform designed for edge-based processing.
Hellbender’s hardware solution didn’t just address each of the challenges above, however. By leveraging its expertise in edge-AI and computer vision, Hellbender designed and built a solution that lowered processing costs from $0.43/km to $0.04/km, drastically improving the economics of crowd-sourced mapping. To do this, the system uses the Intel Movidius chipset to perform feature extraction – like detecting street signs and construction zones – directly on device, eliminating costly cloud computing fees.
Key Features
- Stereo Vision Architecture: Provides reliable depth perception and object recognition.
- On-Device AI Processing: Powered by Intel Movidius for real-time feature extraction.
- High-Precision GPS: L1/L5 integration for accurate geospatial data.
- Connectivity: LTE and Wi-Fi for seamless data transmission.
- Ruggedized Design: Built for durability and manufactured entirely at Hellbender’s US-based facilities.
Impact
By partnering with Hellbender, Hivemapper was able to continue building out their business model on top of a hardware platform that delivers unmatched performance and efficiency for high-precision crowd-sourced mapping. Hellbender’s expertise in edge-AI, optics, and manufacturing has produced a solution that makes high fidelity, up-to-date maps an economically viable reality.
