This week’s roundup of data news from October 3 to October 9, 2026, features the testing of autonomous robotic arms for satellite servicing in orbit and an AI mapping system that combines field observations, drone imagery, and satellite data to monitor Antarctica’s changing ecosystems.
1. Servicing Satellites
Seattle-based robotics startup Orbital Robotics has built autonomous robotic arms that NASA will test aboard the International Space Station using its free-flying Astrobee robot. The system lets Astrobee approach, capture and manipulate satellite-like targets, demonstrating how autonomous servicing could work in microgravity. The test aims to highlight how free-flying robots could eventually repair, upgrade or reposition satellites without astronauts.
2. Explaining Voting Ballots
Colorado-based company The KyleGroup has created a prompt to help voters interpret complex ballot measures. The method starts with official ballot language, which users copy into an AI chatbot and ask it to provide a plain-English explanation and arguments for and against. The approach could make lengthy ballot materials easier to understand by giving voters a clearer way to weigh the potential effects and arguments surrounding each measure.
3. Personalizing Assistants
New York-based startup Persona has launched a personal AI assistant built around a wearable band. The device activates when a user presses a button or flicks their wrist, then sends encrypted data to Persona’s cloud model, which can order food, book travel, take notes, and answer messages. The assistant aims to handle routine tasks on users’ behalf and become more personalized over time by learning their preferences and needs.
4. Speeding Up Clinical Trials
The Department of Health and Human Services has announced a program and three projects to accelerate U.S. clinical trials using AI tools. The effort includes SURPASS, which would simulate trials, analyze incoming data, and automate startup tasks; STACK, which aims to expand trial sites; COMMONS, which would build national research-data infrastructure; and CINCH, which would help patients identify relevant cancer trials.
5. Building Browser Games
Google has launched an AI-powered game-creation platform called Playground that lets users build browser-based games through text prompts instead of coding. Creators can choose a genre, 2D or 3D visuals, and game mechanics, while uploading images that the platform converts into game assets. The AI-powered platform aims to make game development more accessible by helping users turn ideas and visual concepts into playable games without traditional coding.
6. Catching Transplant Issues
Researchers at New York University have trained AI models to detect signs of heart-transplant rejection by analyzing electrocardiograms alongside blood-test biomarkers. The system learned patterns from 5,300 EKG readings of 2,357 transplant recipients and compared them with biopsy results. In a test of 38 patients, the combined model correctly identified 94 percent of patients who were not experiencing rejection, suggesting it could reduce unnecessary invasive biopsies.
7. Building Biology Data
The U.S. government has partnered with Google, Meta, and the nonprofit Biohub to build large open biological datasets for training AI models. The dataset initiative will collect standardized measurements of how cells respond to different conditions, creating data that researchers can use to build predictive models of cellular behavior. The effort aims to accelerate biological research and drug development by making large datasets available to researchers.
8. Mapping Antarctica’s Ecosystems
Researchers at the University of Wollongong in Australia have developed an AI mapping system to monitor Antarctica’s scattered ecosystems. The system combines field observations, drone imagery, and satellite data, allowing researchers to map vegetation from small patches of moss to broader landscape patterns. This approach aims to support long-term biodiversity monitoring by extending detailed field observations across areas that are difficult and expensive to visit.
9. Teaching New Skills
Seattle-based AI startup Yoodli has expanded its AI platform from speech and interview practice into broader workplace training. Employees can rehearse conversations with AI personas, generate training courses from a prompt and receive coaching inside tools, such as Zoom and Teams. The platform also gives managers reports on whether employees’ skills are improving, helping organizations train workers across industries with customizable AI instructors.
10. Detecting Skin Cancer
Clinicians at St Luke’s Hospital in Bradford, England, are using an AI system to speed up assessments of suspicious skin lesions. The tool analyzes three photographs in minutes and helps identify cases that can be discharged or referred for specialist review, while clinicians double-check every result. The hospital says the system has increased patients seen per session from 24 to 32 and reduced unnecessary biopsies by 10 percent.
