Instawork Launches Instacore, a Wearable Camera System for Gig Workers to Train Robots
Instawork, a gig labor platform with 10 million workers, is launching Instacore, a five-camera wearable system that records workers performing tasks to generate training data for robotics companies and AI labs. The system represents a new revenue stream for the company and reflects growing demand for real-world data as AI companies scale robotics development. The initiative raises questions about how gig workers will be compensated for their data and the implications of turning human labor into machine-learning training material.
Instawork, a labor marketplace platform best known for matching hourly workers with shifts at hotels, warehouses, and stadiums, is launching Instacore, a wearable camera system designed to collect real-world data for training robots. The system consists of five cameras mounted on the head, chest, and wrists, connected to a compute backpack weighing less than three pounds and designed to last an eight-hour shift. Workers wearing the system record themselves performing commercial tasks—from food preparation to shelf stocking—and that footage is provided to robotics companies and AI research labs developing machines for physical-world environments. The move reflects a broader industry scramble for training data as major AI companies including OpenAI, Nvidia, Meta, and Tesla accelerate robotics development. Instawork's 10 million workers provide access to diverse real-world tasks and settings, making the data particularly valuable for training robots to handle unpredictable commercial environments. The company, which has raised over $150 million in funding, declined to name specific customers but stated it is working with leading research labs.
What's missing
The article does not disclose whether or how Instawork Pros will be compensated for providing their data and video footage, or what data privacy protections and consent mechanisms are in place for workers whose images and movements are being recorded and shared with third parties.
What different sources said
- Business InsiderLeft
First look: This weird wearable device turns human workers into robot data collectors
Related
Event-Based Scheimpflug LiDAR System Achieves Ultra-Fast 3D Rangefinding
Researchers have developed eSCHORTY, a novel LiDAR system that combines event-based sensors with modulated continuous-wave lasers to generate dense 3D point clouds at over one million megaevents per second. The system addresses limitations of traditional frame-based ranging by providing intrinsic background rejection and operating effectively in high-throughput and cluttered environments. The technology's Scheimpflug geometry enables scalable applications ranging from millimeter to kilometer-scale measurements.
Researchers Develop Multi-Channel Optical Neural Network for Vision Tasks
Scientists have created an optical neural network that uses spatial multiplexing to process multiple inputs simultaneously while serving as a trainable representational coordinate system. The system combines a programmable free-space optical processor with a digital transformer for tasks like image classification and image captioning. This work demonstrates that optical computing can effectively handle complex vision tasks with over one million trainable parameters.
Single-Step Lithography Process Simplifies Silicon Nitride Photonic Chip Manufacturing
Researchers have developed a single-step lithographic process for creating thermo-optic phase shifters on silicon nitride photonic chips, integrating titanium nitride heaters and aluminum interconnects simultaneously. Conventional manufacturing requires multiple separate lithographic steps for heaters, transmission lines, and contacts, making the new approach more efficient. This advancement could accelerate production of reconfigurable photonic integrated circuits used in optical signal processing and quantum computing applications.