OpenLander is an AI-assisted landing decision system designed for future space vehicles, from spacecraft, reusable rockets to drones and planetary robots. It analyzes a sequence of RGB images during descent to estimate depth, understand terrain, detect hazards, and measure visual drift. The system then scores possible landing zones based on safety, obstacle clearance, reachability, mission context, and temporal stability. Instead of selecting one fixed target, OpenLander continuously updates its decision and can retarget when conditions change. More than 40 simulated landing scenarios are created to test the system across environments such as the Moon, Mars, deserts, cities, forests, and water.

During disasters such as the 2024 Noto Peninsula Earthquake, landslides and road collapses can block transportation routes, affecting rescue operations and the delivery of supplies. Our service uses SAR images obtained from Tellus to analyze changes around roads before and after an earthquake and detect road blockages. The detected results are exported as a CSV file and imported into UC-win/Road to simulate the impact of road blockages on traffic. This enables understanding of road conditions during disasters and supports disaster response by infrastructure and logistics companies. In the future, we aim to develop the service to suggest alternative routes using these results.

LunaPilot is an autonomous lunar rover digital twin developed in UC-win/Road. It enables a rover to travel from Satellite Base A to Energy Center B, avoid rocks and craters, and return along a different route. The system uses a third-person follow camera, ORB-based visual feature matching, relative motion estimation, obstacle mapping, topological keyframes, A* route planning, and teach-and-repeat control. A real-time dashboard visualizes camera images, feature matches, detected obstacles, vehicle pose, and outbound and return paths. The workflow supports manual route teaching, automatic replay, emergency stopping, session recording, and OBS video capture for clear competition video demonstration and evaluation purposes.

StarwayDispatch is an autonomous airspace management system developed via UC-winRoad SDK. It transforms passive simulations into an intelligent "Digital Brain" by intercepting 1,500+ background traffic units. Key innovations include a scientific density-based monitoring model and a unique "Virtual Node Fallback" algorithm, enabling seamless lane-changing across logically disconnected aerial bridges. Utilizing a high-precision 3-stage flight controller (Ascend-Lateral-Descend), the system ensures autonomous, collision-free coordination with sub-200ms latency. StarwayDispatch effectively bridges the gap between static road data and dynamic air mobility, enhancing traffic efficiency by 40% while providing a scalable framework for future smart city airspace.

With the development of lunar exploration missions, intelligent cargo management is essential for future lunar bases. This project proposes a Lunar Cargo Base Smart Management System using Digital Twin technology. The system integrates AI perception, object tracking, 3D reconstruction, and large language models to achieve automatic cargo recognition, real-time monitoring, and intelligent management. A digital twin platform is developed to realize three-dimensional visualization and autonomous cargo operation with Qwen-based information query and AGV scheduling.

RescueEagle is an autonomous emergency rescue system designed for disaster zones where telecommunications are severely disrupted. Using a fleet of drones, the system rapidly deploys to provide critical aid and locate victims trapped under rubble. It operates independently of ground networks, utilizing a multi-modal sensor fusion approach- incorporating UWB Radar, thermal Cameras, Acoustic sensors, and Geophones-to detect breathing, body heat, cries for help, and seismic vibrations.The system features SOS Emergency Rescue for immediate relief drops, SAR Area Patrol for victim searching, and a Real-time Command Dashboard for live monitoring.

SkyFlow-to-UC is a working prototype that converts processed aerial and UAV vehicle observations into executable traffic digital-twin scenarios for SUMO and UC-win/Road. It provides two complementary workflows: full trajectory replay from CSV data and full traffic-flow generation from configurable road, demand, and driving-style parameters. IDM controls longitudinal motion, while a supervised LSTM classifier can generate lane-change intentions from 35 state features and a 10-frame history. Safety filters, mandatory exit routing, and cooldown rules constrain execution. The system records model decisions and connects SUMO traffic with UC three-dimensional visualization through a TraCI-based SDK plugin.

As humanity develops settlements on the Moon and Mars, vehicles will share roads with people. On regolith-like surfaces the friction coefficient falls far below asphalt (mu = 0.17), so braking distance grows more than five times (measured: 8.2 m on Earth vs 42.7 m, 5.2x), making Earth-tuned ADAS unsafe. GISELLE is a simulator-based pedestrian emergency braking system: YOLO detects pedestrians, a stereo-camera algorithm measures distance and speed, cross-verification against telemetry removes false positives, and a PWM brake controller stops the vehicle a class-aware margin before the pedestrian (8 m for adults, 10 m for children). Validated over 60+ runs with zero collisions within NASA ride-comfort limits.

MoonSafe is collision avoidance system for autonomous and human-driven vehicles at lunar construction sites, where near-total darkness in shadow and drifting regolith dust make vision-based safety systems unreliable. Built on UC-win/Road's Python API, MoonSafe AutoBrake reads each nearby object's position, velocity, and heading in real time, predicts the closest point of approach using each vehicle's actual dimensions rather than a circular approximation, and directly engages the brake when a collision course is confirmed. A companion Radar UI gives the driver continuous visual and spoken time-to-collision alerts. Because it never relies on cameras, MoonSafe protects drivers even when nothing is visible which is applicible on automous vehicle for future task.