In this paper, we propose a model-free volumetric Next Best View (NBV) algorithm for accurate 3D reconstruction using a Markov Chain Monte Carlo method for high-mix-low-volume objects in manufacturing. The volumetric information gain based Next Best View algorithm can in real-time select the next optimal view that reveals the maximum uncertainty of the scanning environment with respect to a partially reconstructed 3D Occupancy map, without any priori knowledge of the target. Traditional Occupancy grid maps make two independence assumptions for computational tractability but suffer from the overconfident estimation of the occupancy probability for each voxel leading to less precise surface reconstructions. This paper proposes a special case of the Markov Chain Monte Carlo (MCMC) method, the Gibbs sampler, to accurately estimate the posterior occupancy probability of a voxel by randomly sampling from its high-dimensional full posterior occupancy probability given the entire volumetric map with respect to the forward sensor model with a Gaussian distribution. Numerical experiments validate the performance of the MCMC Gibbs sampler algorithm under the ROS-Industry framework to prove the accuracy of the reconstructed Occupancy map and the completeness of the registered point cloud. The proposed MCMC Occupancy mapping could be used to optimise the tuning parameters of the online NBV algorithms via the inverse sensor model to realise industry automation.

Active vision systems with independent eye motion are required to update extrinsic parameters continuously to perform stereo reconstruction. The stereo parameters could be calculated through forward kinematics model plus head-eye parameters of each eye. The head-eye parameters could be calibrated through traditional head-eye calibration method separately. However, head-eye calibration method could not reduce the impact of the kinematics model errors, which may cause reconstruction result unacceptable. Instead, we propose an integrated two-pose calibration method to calculate head-eye parameters of both eyes simultaneously to reduce the impact of model error by deriving two new equations from two poses of the robotic bionic eye. The implemented algorithm has been validated on GAZEBO and our robotic bionic eye. The results demonstrate that the head-eye parameters obtained by our method are more accurate and more robust to kinematics model errors. We have also analyzed the reason why the proposed method is better.

Three-dimensional (3D) triangulation based on active binocular vision has increasing amounts of applications in computer vision and robotics. An active binocular vision system with non-fixed cameras needs to calibrate the stereo extrinsic parameters online to perform 3D triangulation. However, the accuracy of stereo extrinsic parameters and disparity have a significant impact on 3D triangulation precision. We propose a novel eye gaze based 3D triangulation method that does not use stereo extrinsic parameters directly in order to reduce the impact. Instead, we drive both cameras to gaze at a 3D spatial point P at the optical center through visual servoing. Subsequently, we can obtain the 3D coordinates of P through the intersection of the two optical axes of both cameras. We have performed experiments to compare with previous disparity based work, named the integrated two-pose calibration (ITPC) method, using our robotic bionic eyes. The experiments show that our method achieves comparable results with ITPC.

Most of the traditional information-based unknown maze traversal techniques are based on trial and error, which are time-consuming and may trap the robot in infinite loops or dead ends. To address the above issues, this letter proposes a gap based approach for imperfect-unknown mazes traversal named Follow the Nearest-to-final goal and Unvisited Gap (FNUG) for the first time. The algorithm is computationally efficient, and has the ability to deal with loops and dead ends through the revisit check. The main part of this approach is finding out gaps around the robot by analyzing the depth scans, building a coordinate system based topological map in the form of a tree. Then a heuristic selection of the robot’s unvisited-neighbor gap for navigation based on its Euclidean distance from the final target location. Finally Dijkstra’s algorithm is used to find a path from robot’s location to the sub goal. Compared to existing sensor information-based methods, our proposed algorithm not only takes into account the current sensor data, the shape and size of the robot, but also all the sub goals visited so far. This is the key factor in this algorithm that prevents the robot from falling into a dead end or trapped in a loop. To prove the effectiveness of this algorithm, simulations and physical experiments were performed using using a 2-wheeled differential mobile robot.