Nuvonix partnered with Condor AI to bring integrated Edge AI solutions to enterprise customers. The offering combines edge AI cameras, AI gateways, compute hardware, and KALKI, Condor AI’s platform for orchestrating AI workloads and devices. Target applications include video analytics, fleet management, industrial operations, and security, helping organizations move beyond pilots toward scalable deployments. [link]

Oxipital AI launched a foreign-material detection capability for high-speed food production lines. Using its existing OX and VX hardware platforms, the system identifies contaminants such as plastic, wood, bone, and personal protective equipment in real time. The company reports detection accuracy above 99%, alongside automated responses, audit logs, and analytics through its V-CORTX platform. [link]

Maxinsights raised $30 million in seed funding, led by Peak XV, to expand real-world training data for robotics models. Its network of 5,000+ contractors across six countries captures first-person footage of everyday tasks. The company also introduced MaxLoop for data-quality checks and MaxGlove for grip-force measurement, addressing growing demand for multimodal physical AI training data. [link]

Peregrine.ai is joining the YellowFox group after more than three years of collaboration. Its computer vision technology analyzes driving behavior, traffic situations, vehicle surroundings, and cabin activity in real time, helping fleet operators identify hazards and document incidents. The Peregrine.ai brand, technology offering, Berlin office, and existing customer relationships will continue under the expanded partnership. [link]

AI startup Elorian launched Glimpse, a visual reasoning model available through API access and a chat interface. Unlike systems that rely on a single image interpretation, Glimpse can revisit images, zoom into regions, extract video frames, and use code to measure visual details. The company targets robotics, scientific research, media, and other video-intensive applications. [link]

Pony.ai, inDrive, and NVIDIA have signed an MoU to explore autonomous ride-hailing in Kazakhstan, with Astana as the proposed launch city. Pony.ai would supply Gen-7 robotaxis, inDrive would operate the ride-hailing platform, and NVIDIA would provide AI infrastructure, including Hyperion. The project will assess technical, operational, regulatory, and commercial feasibility. [link]

Ambient.ai secured a Series C round led by Allegion Ventures, which made its second strategic investment in the company. Its AI platform transforms existing cameras and sensors into proactive security systems, powered by edge-optimized vision-language models. Ambient reported 90% annual recurring revenue growth in fiscal 2026 and cash-flow positivity during the first half of the year. [link]

Eka Robotics raised $150 million in Series A funding, co-led by Andreessen Horowitz and F-Prime, to scale its general-purpose robotics technology. Its vision-force-action model combines vision, physical interaction, simulation, and reinforcement learning to improve dexterous manipulation. The company reports robots performing up to 1,500 operations per hour across varied tasks, targeting faster and more adaptable real-world automation. [link]

Mobileye received TÜV SÜD’s first inspection certificate for its Automatic Emergency Braking development process, confirming conformity with ISO 21448 (SOTIF). The independent assessment reviewed how Mobileye identifies and mitigates risks arising from sensor limitations, perception systems, and real-world conditions. The milestone underscores the importance of systematic safety validation in advanced driver-assistance technology. [link]

A new study, Probing Negation Understanding in Vision-Language Models With Object Attention, reveals that six VLMs struggle with negated prompts, consistently underperforming on negation-based questions. Comparing model attention with human eye-tracking patterns, researchers find that attention shifts rarely mirror human reasoning, highlighting critical challenges for reliable visual understanding in safety-critical applications. [link]

Researchers introduce OLiVES, an outdoor low-light video benchmark featuring 25,254 aligned normal- and low-light frames and 46,366 segmentation annotations. Captured using cinema-grade cameras, it includes dynamic scenes, wildlife, occlusions, and complex textures. Experiments show that training on OLiVES improves video enhancement and segmentation, advancing computer vision for surveillance, robotics, and wildlife monitoring.. [link]

In the latest Beyond Fulfilment podcast, Gather AI CEO Sankalp Arora shares his journey from near quitting to raising $74M. He discusses customer discovery, warehouse automation, autonomous drones, leadership, and building resilient teams that deliver measurable impact.

In the latest Inside The Lane episode of Our Parking Industry Podcast, Automotus CEO Jordan Justus discusses how computer vision and automated curb management help cities and airports reduce congestion, improve enforcement, optimize curb space, and develop data-driven transportation policies.

In an interview with EE Times Asia, Deepx CEO Lokwon Kim explains how open software ecosystems, low-power AI accelerators, and edge deployment can accelerate physical AI adoption across intelligent vision, robotics, industrial automation, and connected devices.