Ability Of Autonomous Navigation Research Articles

Black Titania Janus Mesoporous Nanomotor for Enhanced Tumor Penetration and Near-Infrared Light-Triggered Photodynamic Therapy.

Thanks to their excellent photoelectric characteristics to generate cytotoxic reactive oxygen species (ROS) under the light-activation process, TiO2 nanomaterials have shown significant potential in photodynamic therapy (PDT) for solid tumors. Nevertheless, the limited penetration depth of TiO2-based photosensitizers and excitation sources (UV/visible light) for PDT remains a formidable challenge when confronted with complex tumor microenvironments (TMEs). Here, we present a H2O2-driven black TiO2 mesoporous nanomotor with near-infrared (NIR) light absorption capability and autonomous navigation ability, which effectively enhances solid tumor penetration in NIR light-triggered PDT. The nanomotor was rationally designed and fabricated based on the Janus mesoporous nanostructure, which consists of a NIR light-responsive black TiO2 nanosphere and an enzyme-modified periodic mesoporous organosilica (PMO) nanorod that wraps around the TiO2 nanosphere. The overexpressed H2O2 can drive the nanomotor in the TME under catalysis of catalase in the PMO domain. By precisely controlling the ratio of TiO2 and PMO compartments in the Janus nanostructure, TiO2&PMO nanomotors can achieve optimal self-propulsive directionality and velocity, enhancing cellular uptake and facilitating deep tumor penetration. Additionally, by the decomposition of endogenous H2O2 within solid tumors, these nanomotors can continuously supply oxygen to enable highly efficient ROS production under the NIR photocatalysis of black TiO2, leading to intensified PDT effects and effective tumor inhibition.

ST$^{2}$: Spatial-Temporal State Transformer for Crowd-Aware Autonomous Navigation

Empowering an intelligent agent with the ability of autonomous navigation in complex and dynamic environments is an important and active research topic in embodied artificial intelligence. In this letter, we address this challenging task from the view of exploiting both the spatial and temporal states of a mobile robot interacting with the crowded environment. Specifically, we propose a Spatial-Temporal State Transformer (ST <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{2}$</tex-math></inline-formula> ) to encode the states while leveraging the deep reinforcement learning method to find the optimal navigation policy accordingly. Technically, the proposed ST <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{2}$</tex-math></inline-formula> model consists of a global spatial state encoder and a temporal state encoder, which are built upon the Transformer structure. The spatial state encoder is devised to extract the global spatial features and capture the spatial interaction between pedestrians and the robot. The temporal state encoder is designed to model the temporal correlation among consecutive frames and infer the dynamic relationship of the spatial position transformation. Based on the comprehensive spatial-temporal state representation, the value-based reinforcement learning method is leveraged to obtain the optimal navigation policy. Extensive experiments demonstrate the superiority of the proposed ST <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^{2}$</tex-math></inline-formula> over representative state-of-the-art methods. The source code will be made publicly available.

Monocular Visual-Inertial Sensing of Unknown Rotating Objects: Observability Analyses and Case Study for Metric 3D Reconstructing of Space Debris

The fusion of a monocular camera and an inertial sensor (i.e., IMU) can help the autonomous navigation ability of a mobile robot, as well as obtain the environmental information. However, it is difficult to adapt the fusion for dynamic objects. Although several recent studies have tracked or even retrieved simple structural information about dynamic objects, there is no reported visual-inertial sensing research for unknown rotating objects, especially for those that are not surrounded by any static environmental visual feature. In fact, such rotating objects indeed represent a class of real and challenging scenarios, e.g., space debris. This letter discusses the monocular visual-inertial sensing of unknown rotating objects. Our first contribution is the comprehensive nonlinear observability analyses. We theoretically prove that the monocular visual-inertial sensing possesses favourable observability properties with respect to rotating objects, and we derive all observable and unobservable modes for the cases with unknown gravity and without gravity. The second contribution of this letter is a validation case study, i.e., the metric 3D reconstruction of space debris by means of a micro space robot. We propose a practical inertial sensor-assisted Structure from Motion strategy that can perceive rotating space debris. The experiment demonstrates that by using only the onboard monocular camera and IMU, the robot can obtain sufficient information about the space debris, such as all rotation parameters and even a dense point cloud model with absolute metric scale information. This case study confirms our theoretical results.

Design of UAV Flight Control System

In order to improve the control performance of UAV. For a small UAV, the hardware design scheme of UAV flight control computer with STM32 as the processing core is proposed and implemented. The overall scheme design, specific hardware selection and interface design of the system are given in detail. STM32, as the main controller, expands the parallel A/D, D/A, asynchronous serial communication interface and memory around, enriches the system hardware interface resources, and selects the peripheral expansion circuit with complex programmable logic controller (CPID) to realize interrupt control. This flight control computer has the characteristics of small volume and strong autonomous navigation ability.

Observability of a class of under-observation systems

Exploring the deep space of the universe is crucial to promote research on basic scientific issues, such as the origin of the solar system, evolution of planets, and origin of life. Using probes to detect deep-space objects has become the focus of the development of various aerospace powers. As the deep-space exploration task is performed far away from the earth, it is difficult to guarantee the safe and reliable operation of deep-space probes by relying on ground control networks to guide their navigation; as such, deep-space probes should have autonomous navigation ability. An autonomous navigation system based on sequence images has the advantages of low cost, low power consumption, full application, and simple data structure, which can effectively reduce the space and energy burden of the autonomous navigation system of deep-space probes, and is an ideal means to realize autonomous navigation in deep-space probes. However, because of the lack of prior information about deep-space exploration targets and systematic error in a deep-space complex environment, the autonomous navigation system based on sequence images is a typical under-observable system. To improve the accuracy and engineering practicability of typical autonomous navigation systems, it is necessary to guide the design and engineering realization of autonomous navigation systems based on theoretical research on observability. In response to the actual needs of observability theory research, in this paper, we first systematically expound the current research status of existing theories and methods from three aspects: characterization, determination, and quantification of the observability of nonlinear systems. Then, the observability analysis method of a class of under-observation systems—autonomous navigation systems based on sequence images—is investigated. The research status of observability for such systems is given, and the advantages and disadvantages of each method are discussed. Finally, by summarizing existing results, limitations in the observability research of under-observation systems at present and future development trends are discussed.

Path planning for active SLAM based on deep reinforcement learning under unknown environments

Autonomous navigation in complex environment is an important requirement for the design of a robot. Active SLAM (simultaneous localization and mapping) combining, which combine path planning with SLAM, is proposed to improve the ability of autonomous navigation in complex environment. In this paper, fully convolutional residual networks are used to recognize the obstacles to get depth image. The avoidance obstacle path is planned by Dueling DQN algorithm in the robot’s navigation; at the same time, the 2D map of the environment is built based on FastSLAM. The experiments show that the proposed algorithm can successfully identify and avoid moving and static obstacles with different quantities in the environment, and realize the autonomous navigation of the robot in a complex environment.

Autonomous celestial navigation method of asteroid probe based on angle measurement and velocity measurement

Autonomous navigation is one of the key technologies to ensure the successful implementation of deep space exploration mission. Aiming at the requirement of celestial autonomous navigation ability of asteroid probe in practical engineering tasks, the integrated autonomous navigation method combined celestial spectral velocity measurement with celestial image angle measurement is proposed to realize continuous autonomous, real-time and high-precision navigation of asteroid probe. Considering the limitation of traditional radio navigation on the ground and the high requirement for autonomous survivability of asteroid probe, the task of asteroid explorer has an urgent need for real-time and autonomous navigation ability. The current autonomous navigation method mainly extracts the angle measurement information of the probe relative to the target celestial planet from the optical image information of the target celestial planet, and then calculates and determines the navigation state of the probe. However the accuracy of velocity estimation by this navigation method is limited, which still cannot fully meet the current demand of asteroid probe. In view of the shortcomings of the main existing autonomous navigation methods, this paper proposes a direct velocity measurement method based on astronomical spectrum, and combines it with the astronomical angle navigation method to form the integrated autonomous navigation method. Based on the integrated autonomous navigation method, the navigation system model and filtering algorithm are derived, and the observability and navigation accuracy performance of the integrated navigation system are analyzed. Taking the Ceres exploration mission as an engineering background, the simulation results show that the stellar velocity information measured by astronomical spectroscopy can greatly improve the integrated navigation performance. And the accuracy of position and velocity estimation results of the probe can restrain the influence of errors more effectively and enhance the reliability of navigation system. Compared with the traditional terrestrial radio navigation or angle measurement navigation methods, the integrated navigation results have higher accuracy and better real-time performance, which can provide accurate navigation information for asteroid probe orbit modification. The integrated navigation method derived in this paper is effective, reliable and easy to implement. It provides a reference for the implementation of the asteroid exploration project and the subsequent major deep space missions in China.

Comparing satellite orbit determination by batch processing and extended Kalman filtering using inter-satellite link measurements of the next-generation BeiDou satellites

The next-generation BeiDou satellites are equipped with inter-satellite link (ISL) payloads to realize the ability of autonomous navigation (Auto-Nav). We focus on the application of batch processing and extended Kalman filtering (EKF) algorithm on Auto-Nav of BeiDou satellite system (BDS). The mathematical model of dual one-way measurements and principle of batch mode and EKF are introduced. Using real ISL measurements, Auto-Nav experiments are conducted with batch processing and EKF, respectively. For the sub-constellation, the EKF with a priori equipment delay constraints is proposed. The result shows that (1) with three ISLs and only one anchor station, the ISL measurements are sparsely distributed and the coverage of the whole arc is about 31%. The observations and dynamic models per epoch contribute more to satellite position and velocity parameters than to equipment delay parameters. (2) For batch processing, the overlap precision of precise orbit determination (POD) with ISLs and ground-satellite links (GSLs) is about 0.1 m in the radial direction and is better than 1 m three-dimensionally. The variation of the estimated equipment delays is within $$\pm \,0.6$$ ns. The observation residuals of ISLs behave such as a normal distribution, while the residual of GSLs show periodical variation due to uncorrected troposphere delay. (3) For the EKF, the ISL-only orbit determination is sensitive to the accuracy of the initial state. Compared with batch result, the precision of ISL-only orbit determination using EKF is better than 2 m given accurate initial states. The filtering does not show constellation drift or rotation within 8 days. However, with approximate initial states which position accuracy is 100 m, the precision of POD decreases to dozens of meters. (4) For the EKF, the accuracy of POD improves to 1.5 m in three dimensions with the support of an anchor station. The period for equipment delay parameters to converge is about 24 h. However, the convergence rate of equipment delay parameters is much slower than that of satellite state parameters. The possible reason is the unbalanced contribution of observation and dynamic model information on the estimated parameters. Thus, it is better to constrain the equipment delay parameters with a priori information while filtering.

Camera Recognition and Laser Detection based on EKF-SLAM in the Autonomous Navigation of Humanoid Robot

The ability of autonomous navigation of the humanoid robot under unknown environment is very important to real-life applications. EKF-SLAM based on the camera recognition and laser detection for humanoid robot NAO is presented in this paper. Camera recognition is used to recognize if the object is a landmark. Because the computational resources needed for the feature-based position estimation are quite expensive, the laser instead of the camera provides the position of the landmark. A fractional order proportional-integral (PI) controller is designed to reduce the derivation of the NAO robot from the desired path during autonomous navigation. Experiments show that the proposed method is valid and reliable for autonomous navigation of the NAO robot under unknown environment.

Performance Enhancement of a USV INS/CNS/DVL Integration Navigation System Based on an Adaptive Information Sharing Factor Federated Filter.

To improve the ability of autonomous navigation for Unmanned Surface Vehicles (USVs), multi-sensor integrated navigation based on Inertial Navigation System (INS), Celestial Navigation System (CNS) and Doppler Velocity Log (DVL) is proposed. The CNS position and the DVL velocity are introduced as the reference information to correct the INS divergence error. The autonomy of the integrated system based on INS/CNS/DVL is much better compared with the integration based on INS/GNSS alone. However, the accuracy of DVL velocity and CNS position are decreased by the measurement noise of DVL and bad weather, respectively. Hence, the INS divergence error cannot be estimated and corrected by the reference information. To resolve the problem, the Adaptive Information Sharing Factor Federated Filter (AISFF) is introduced to fuse data. The information sharing factor of the Federated Filter is adaptively adjusted to maintaining multiple component solutions usable as back-ups, which can improve the reliability of overall system. The effectiveness of this approach is demonstrated by simulation and experiment, the results show that for the INS/CNS/DVL integrated system, when the DVL velocity accuracy is decreased and the CNS cannot work under bad weather conditions, the INS/CNS/DVL integrated system can operate stably based on the AISFF method.

An Intelligent Robotic Hospital Bed for Safe Transportation of Critical Neurosurgery Patients Along Crowded Hospital Corridors.

We present a novel design of an intelligent robotic hospital bed, named Flexbed, with autonomous navigation ability. The robotic bed is developed for fast and safe transportation of critical neurosurgery patients without changing beds. Flexbed is more efficient and safe during the transportation process comparing to the conventional hospital beds. Flexbed is able to avoid en-route obstacles with an efficient easy-to-implement collision avoidance strategy when an obstacle is nearby and to move towards its destination at maximum speed when there is no threat of collision. We present extensive simulation results of navigation of Flexbed in the crowded hospital corridor environments with moving obstacles. Moreover, results of experiments with Flexbed in the real world scenarios are also presented and discussed.

Autonomous orbit and attitude determination including time prediction based on XNAV and ultraviolet sensor

为提高深空探测器的自主导航能力, 利用脉冲星导航的脉冲到达时间和脉冲星 角位置测量值、紫外敏感器中心天体质心相对于探测器的方向矢量和距离测量 值以及紫外敏感器输出的航天器姿态角, 以探测器在惯性坐标系下的位置和速度、 探测器本体坐标系相对于惯性坐标系的姿态角、星载时钟钟差为系统状态变量, 通过联邦扩展卡尔曼滤波器估计组合导航系统的系统状态, 并利用火星环绕段 轨道数据进行仿真实验. 仿真结果表明, 该组合导航方法能够使火星轨道器 在环绕段飞行中同时进行定轨、定姿和授时, 且具有较高的导航精度和授时能力.

Vision-based object detection and tracking for autonomous navigation of underwater robots

Underwater robots have been an emerging research area being at the intersection of the field of robotics and oceanic engineering. Their applications include environmental monitoring, oceanographic mapping, and infrastructure inspections in deep sea. In performing these tasks, the ability of autonomous navigation is the key to a success, especially with the limited communications in underwater environments. Considering the highly dynamic and three-dimensional environments, the autonomous navigation technologies including path planning and tracking have been one of the interesting but challenging tasks in the field of study. Cameras have not been at the center of attention as an underwater sensor due to the limited detection ranges and the poor visibility. Use of visual data from cameras, however, is still an attractive method for underwater sensing and it is especially effective in the close range detections. In this paper, the vision-based object detection and tracking techniques for underwater robots have been studied in depth. In order to overcome the limitations of cameras and to make use of the full advantages of image data, a number of approaches have been tested. The topics include color restoration algorithm for the degraded underwater images, detection and tracking methods for underwater target objects. The feasibilities of the proposed algorithms have been demonstrated in the experiments with an underwater robot platform and the results have been analyzed both qualitatively and quantitatively.