Synthetic Dataset Generation for 3D SLAM Evaluation

Rosbags recorded from a simulation environment, containing data from a simulated Velodyne(VLP16), IMU, camera RGB, wheel odometry and robot's real position. Each sequence contains different challenges.TopIcs: /ground_truth/odom: topic that publishes robot's real pose in odometry message format. /scan: topic that publishes the 3D Lidar Point Cloud. /clock: topic that publishes simulation time, for synchronization. /odom: topic that publishes wheel Odometry. /imu/data: topic that publishes the full raw IMU information including orientation (as a quaternion), angular velocity, and linear acceleration, which is essential for SLAM and sensor fusion. /imu/rpy: topic that publishes the same orientation but converted to roll, pitch, and yaw angles. /camera/compressed: topic that publishes the compressed images from robot's RGB camera. /tf: topic that publishes the tf tree. Sequences: Static: this sequence uses the map without additional challenges to seve as the baseline result for comparisson. Time: 706 seconds; Distance Traveled: 530 meters; laps: 2 Dynamic: this sequence introduces moving people within the environment to simulate real-world, dynamic conditions. This setup evaluates the SLAM algorithm's robustness to non-static obstacles and tracking consistency in the presence of human motion. Time: 765 seconds; Distance Traveled: 567 meters; laps: 2 Ramp 15°: this sequence features a scenario where the robot ascend and descend a 15-degree inclined surface. This sequence introduces significant non-planar motion characterized by vertical displacement and pitch rotation, posing a challenge for algorithms not optimized for elevation changes. Time: 713 seconds; Distance Traveled: 540 meters; laps: 2 The Symmetric Corridor sequence includes structural modifications to the environment, specifically the addition of two symmetric walls along the central corridor. This creates a visually and structurally repetitive scene with low-feature variance, which tests the SLAM system's ability to resolve feature ambiguity and maintain accurate localization. The sequence duration is 316 seconds, with a distance of 237 meters over two laps. Time: 316 seconds; Distance Traveled: 237 meters; laps: 2 Ramp 30°: this sequence presents a more difficult incline than the 15° scenario. The robot attempts to ascend a 30-degree slope but fails, inducing wheel slippage. This environment is particularly challenging for odometry-dependent SLAM methods that rely on traction for accurate motion estimation. Time: 802 seconds; Distance Traveled: 583 meters; laps: 2 Slippery Zone: this sequence introduces a low-friction surface designed to deliberately cause wheel slippage. This environment challenges SLAM algorithms that depend heavily on wheel odometry. The sequence duration is 789 seconds, and the robot travels 589 meters. Time: 789 seconds; Distance Traveled: 589 meters; laps: 2