TMC Edge: Thermoeye’s ‘AI’ Camera
Thermoeye, a company specializing in thermal vision solutions, has announced the release of their TMC Edge thermal edge ‘AI’ camera, scheduled for this March. The TMC Edge is a thermal camera with ‘AI’, designed for use in autonomous driving, robots, drones, and other applications. It is noted as the first thermal camera capable of object detection and distance estimation simultaneously.
It provides forward visibility in conditions including rain, snow, night, and fog, which can be challenging for regular RGB cameras and lidar. It features a resolution of 640 × 512 (VGA) and a 6-TOPS ‘AI’ chip that performs object detection and distance estimation up to 150 meters with inaccuracy of less than 2 per cent. Its compact design, with dimensions under 5 cm, allows easy integration into different hardware systems, and the standardized bracket facilitates convenient installation.
The TMC Edge eliminates the need for an additional distance sensor, freeing up space and increasing efficiency in hardware design. It has a high-resolution, high-sensitivity thermal image sensor, offering complete forward visibility even in harsh weather conditions.
Thermoeye is preparing to commercialize the TMC Edge for autonomous applications, and is discussing technology development and cooperation with tier-1 automotive suppliers. According to CEO Dohwi Kim, Thermoeye aims to build a safe autonomous environment based on thermal cameras and is looking to integrate their technology into current and future markets.
FLIR, VSI Labs: Thermal Imaging Helps Meet Nighttime AEB Safety Standard
Teledyne FLIR OEM and VSI Labs have announced test results using the new FMVSS 127 requirements for pedestrian automatic emergency braking (PAEB). VSI Labs’ vehicle, equipped with Teledyne FLIR’s latest AEB thermal camera, passed all tests, unlike three other 2024-model vehicles which failed two or more nighttime scenarios.
FMVSS 127 requires PAEB systems in all new passenger cars and light trucks under 10,000 pounds by September 2029. These systems must detect pedestrians in various lighting conditions and at higher speeds due to pedestrian fatalities, especially at night, which accounted for 77.7 per cent of deaths in 2022.
NHTSA showed that meeting FMVSS 127 requirements is possible, as the 2023 Toyota Corolla Hybrid XLE passed all PAEB tests at maximum speed. However, other tested vehicles failed some nighttime tests due to challenges with visibility and sensor capabilities.
VSI Labs tested a Ford Fusion Hybrid with a thermal-fused PAEB system using Teledyne FLIR’s thermal camera, HD radar, and visible camera against three state-of-the-art 2024 COTS PAEB systems that lacked thermal cameras. The tests revealed :
- Daytime Performance: Both thermal-fused and COTS PAEB systems passed all tests.
- Nighttime Performance: Only the thermal-fused system passed all tests, showcasing better detection regardless of headlight performance.
Thermal cameras can see through darkness, shadows, glare, smoke, and fog, and they can detect wildlife. Tests by VSI Labs indicated current top-rated PAEB systems struggle at night, emphasizing thermal cameras’ potential to meet FMVSS 127 requirements by 2029 for enhanced pedestrian safety. Testing was carried out at the American Center for Mobility in Michigan, using a heated adult pedestrian test mannequin representing a human’s thermal signature.
DVN comments
FLIR cameras utilize far infrared radiation to produce images. At night, objects with higher temperatures, such as pedestrians or animals, as well as vehicle engines or tires, emit more infrared radiation compared to the road and surrounding environment due to increased thermal contrast. However, during some summer days, these thermal contrasts may diminish, potentially hindering object detection. Fog scatters infrared radiation but less than visible or near infrared light, allowing FLIR cameras to detect hidden objects. However, thicker fog increases scattering and reduces FLIR camera effectiveness. When thermal contrasts decrease or in thick fog, it becomes necessary to integrate depth images from radars.