An unsupervised lightweight transformer for depth completion that fuses RGB and sparse depth using depthwise blocks and a cross-hierarchical-attention decoder to improve edge …
An intelligent quadruped control system for extreme terrain that combines a fuzzy-logic high-level controller (MFLC) with a neural network-based low-level controller to improve …
A robust legged-robot state estimator that extends VINS-Fusion by adding a preintegrated foot velocity factor, enabling reliable estimation in outdoor in-the-wild conditions …
A quadruped target-tracking system that jointly plans traversable, obstacle-free local trajectories using SLAM, RGB-D object detection, and QR-SCAN-style trajectory scoring, and …
An unsupervised depth completion method that leverages line features from visual SLAM via constrained Delaunay triangulation and a mesh depth refinement module to produce dense, …
A trajectory planning framework for UAV target tracking in low-light environments, combining thermal and ToF-based perception with minimum-snap trajectory optimization under …
An unconstrained line-based visual SLAM system that leverages vanishing point measurements to resolve line degeneracy and guarantee observability for accurate structural mapping.
A robust UAV autonomous landing system on a moving platform, fusing object detection and UWB measurements with VIO/encoder data via an Extended Kalman Filter to estimate relative …
A lightweight local exploration planner for quadruped robots that uses kinodynamics-aware precomputed trajectories and a two-step cylindrical collision/traversability check (via …
Comprehensive benchmarking of visual(-inertial) odometry algorithms on various NVIDIA Jetson platforms, evaluating accuracy and computational efficiency.