Autonomous Navigation Lab

The 3D view uses projected solid geometry and an orbit camera to show the field, trees, obstacles and robot from different angles. It visualizes the same planar navigation model as the 2D map: obstacle height is illustrative, and the planner does not estimate terrain or plan in three dimensions. LiDAR rays are shown on the field plane. Drag to orbit, scroll or use the camera buttons to zoom, and switch on Follow robot for a closer moving view. Clicking still edits the ground map under the cursor.

A* searches an eight-connected grid at 0.25 m resolution using Euclidean travel costs and a Euclidean heuristic. Occupancy is inflated by a circular robot envelope (radius 0.72 m) plus your selected safety margin. Every grid transition and every line-of-sight shortcut is sampled at intervals no larger than 0.04 m for collision clearance. The robot follows the simplified route by rotating in place, then driving toward each waypoint, with speed limited by its remaining stopping distance. Each executed translation is checked again against the inflated map.

This is a static, fully known map demonstration. The LiDAR overlay is a simulated 360° range scan (48 rays, 5.2 m range); it visualizes nearby geometry and does not construct the planning map. Field edits trigger a new plan immediately. It does not model slip, localization error, dynamic actors, sensing uncertainty, or safety-certified control. Small grid and sampling discretization errors remain; the extra margin is not a certification. The closest-clearance metric is measured from the circular robot envelope to the nearest obstacle or field boundary, including the starting position.

CSV data contains simulation time, robot position, heading, speed, current clearance, travelled distance, goal position, safety margin and plan count at 0.2 s intervals. These are model outputs for education and design exploration, not field-trial measurements.