The invention discloses an ordinary object recognizing method based on 2D and 3D SIFT feature fusion. The ordinary object recognizing method based on 2D and 3D SIFT feature fusion aims to increase the ordinary object recognizing accuracy. A 3D SIFT feather descriptor based on a point cloud model is provided based on Scale Invariant Feature Transform, SIFT (2D SIFT), and then the ordinary object recognizing method based on 2D and 3D SIFT feature fusion is provided. The ordinary object recognizing method based on 2D and 3D SIFT feature fusion comprises the following steps that 1, a two-dimension image and 2D and 3D feather descriptors of three-dimension point cloud of an object are extracted; 2, feather vectors of the object are obtained by means of a BoW (Bag of Words) model; 3, the two feature vectors are fused according to feature level fusion, so that description of the object is achieved; 4, classified recognition is achieved through a support vector machine (SVN) of a supervised classifier, and a final recognition result is given.

Ordinary object recognizing method based on 2D and 3D SIFT feature fusion

The invention discloses a double piezoelectric actuators type micro flapping wing aircraft based on soft hinges. The double piezoelectric actuators type micro flapping wing aircraft comprises two bent beam type twin-lamella piezoelectric actuators, two transmission mechanisms in central symmetry distribution, an outer frame surrounding the periphery of the aircraft, and two wing structures containing driven torsion hinges. The bent beam type twin-lamella piezoelectric actuators are bent under the excitation of voltage, the wings are driven to move by the transmission mechanisms, so that a force needed by the motion of the aircraft is generated. By adopting the structure and using the laser micro-cutting technology and soft hinge connection style, the processing and manufacturing of the micro flapping wing aircraft with wingspan of 3 centimeters are achieved. The mechanism is simple and compact in structure; two actuators are respectively used for controlling the two wings so that the aircraft can achieve other motions besides hovering and straight-line flight.

Double piezoelectric actuators type micro flapping wing aircraft based on soft hinges

The invention discloses a target identification method based on three-dimensional point cloud data. The target identification method comprises the following steps of: carrying out external cuboid obtaining on three-dimensional point cloud data of a marked category; extracting a regular geometry from the three-dimensional point cloud data of the marked category; constructing a model database; constructing a model data table; and inputting point cloud data of a target to be identified and carrying out online identification. A target identification system based on the three-dimensional point cloud data comprises a first module, a second module, a third module, a fourth module and a fifth module, wherein the first module is used for executing external cuboid obtaining on the three-dimensional point cloud data of the marked category; the second module is used for executing extraction of the regular geometry from the three-dimensional point cloud data of the marked category; the third module is used for executing construction of the model database; the fourth module is used for executing construction of the model data table; and the fifth module is used for executing input of the point cloud data of the target to be identified and online identification. The method and the system are convenient and rapid in the target identifying process, and are high in accuracy. The method and the system are widely applied to the technical field of mode identification.

Target identification method and system based on three-dimensional point cloud data

The invention provides a human face depth and surface normal vector predication method based on a dilated convolution neural network The method includes steps of training the dilated convolution neural network S1, constructing the dilated convolution neural network including a plurality of convolution layers, a plurality of dilated convolution layers and a plurality of deconvolution layers that are connected in sequence, wherein each convolution layer is connected with a normalized operation and an motivation operation; S2, initializing the weight value of the dilated convolution neural network; S3, inputting training pictures into the dilated convolution neural network and performing iteration training on the dilated convolution neural network targeting at minimizing a cost function andupdating the weight value after each iteration process; S4, inputting testing pictures into the dilated convolution neural network obtained through training and outputting a corresponding human face depth map and a surface normal vector map; S5, judging whether the predication precision of the dilated convolution neural network obtained through training meets requirements or not according to the output human face depth map and the human face normal vector method, ending the training if the precision meets the requirements, and returning to S3 for training again if the precision does not meetsthe requirements

Human face depth and surface normal vector predication method based on dilated convolution neural network

The invention discloses a structure analysis based identification method for an object in a point cloud scene. The method comprises the steps of step 1: segmenting the point cloud scene by utilizing a regional growth method and a distance-based clustering algorithm, and extracting planes in the point cloud scene based on the property of a Gaussian sphere; step 2: for the extracted planes, representing a single plane by a node, representing a connection relationship between the planes by an edge, analyzing the attributes of the node and the edge, and defining the type of connection between the planes; step 3: analyzing the structure of a target object and recording a connection code of the target object; and step 4: performing plane combination and target identification, and finishing object identification by comparison with the code of the target object. According to the structure analysis based identification method for the object in the point cloud scene, the defects that an existing identification technology is large in calculation amount, incomplete in point cloud data, easy to influence by noises and not fit for identifying an object in point cloud data of a large scene are overcome.

Structure analysis based identification method for object in point cloud scene

The invention discloses a method for identifying objects in 3D point cloud data. 2D SIFT features are extended to a 3D scene, SIFT key points and a surface normal vector histogram are combined to achieve scale-invariant local feature extraction of 3D depth data, and the features are stable and reliable. A provided language model overcomes the shortcoming that a traditional visual word bag model is not accurate and is easily influenced by noise when using local features to describe global features, and the accuracy of target global feature description based on the local features is greatly improved. By means of the method, the model is accurate, and identification effect is accurate and reliable. The method can be applied to target identification in all outdoor complicated or simple scenes.

Method for identifying objects in 3D point cloud data

The invention discloses a piezoelectric ceramic flapping-wing-type robot. The piezoelectric ceramic flapping-wing-type robot comprises a robot body support frame, robot wings, a piezoelectric driver, a driving electric cable, a transmission amplifying mechanism and driving feet, wherein the robot wings are based on a bionic design and are similar to a dipster hoverfly in shape; when the robot walks on land, a piezoelectric ceramic sheet is selected as the piezoelectric driver, and the piezoelectric ceramic sheet and a single-bent-end transmission amplifying mechanism form the driving feet; the driving electric cable inputs two common grounding sine wave signals for respectively driving the two driving feet, and the robot is driven to do controllable plane motion by utilizing a resonant oscillation principle. When the robot flies in the sky, the driving feet are taken down; a piezoelectric ceramic bimorph element is used as the piezoelectric driver to input a driving voltage signal; by a four-connecting-rod transmission mechanism, the oscillation is amplified and is converted into flapping of wings, thereby driving the robot body to fly. The robot is simple, light and convenient, and has certain adaptability to the environment.

Piezoelectric ceramic flapping-wing-type robot

The invention provides a speed control method of a frog board robot. The method comprises that the forward speed when the frog board robot walks on the ground is kept stable by utilizing a fuzzy PID algorithm, and the speed of the robot on the road of relatively large gradient in the free descending sliding process is kept stable by controlling a brake. The problem that the speed of the frog board robot is hard to control due to nonlinearity, hysteresis quality and time variation is solved by utilizing the fuzzy PID algorithm and the break which can realize smooth speed control for the frog board robot, the robot is prevented from jittering and insufficient power in the forward process, and the robot is avoided from danger caused by too high speed in the descending process.

Speed control method of frog board robot

The invention provides a swinging arm driving-type motion control method for an amphibious frog board robot The method comprises specific steps: a leg and arm position theta a of the robot and the speed Vc of the robot are measured; the mass center of the robot in the advancing direction has the accelerated speed ac=K*(Vaim-Vc), wherein Vaim is the aimed speed and K is a preset parameter; theta a, Vc and ac are used to calculate motor control quantity omega according to a built advancing motion control model for the robot on land or in water; and according to the motor control quantity omega, the motor is controlled to drive the robot to advance on land or in water A unified controller can be used for realizing longitudinal speed control on the robot in the land-water environment, complexity of the amphibious robot control algorithm is greatly simplified, and loss of time and energy of the amphibious robot caused by environment detection, execution mechanism switching and control algorithm switching can be avoided

Swinging arm driving-type motion control method for amphibious frog board robot

Accurate terrain estimation is essential for unmanned ground vehicle (UGV) to achieve autonomous navigation. However, some existing approaches are difficult to reconstruct the 3D surface in real-time because of serious amount of computation. A method of environmental terrain estimation based on multi-scale radial basis function with a 3D lidar mounted on a moving vehicle was proposed in this paper. By combining kernel template with terrain grid map, the approach is applicable to the non-uniform density distribution of laser point cloud with the increasing detection range and improves the efficiency of calculation. What's more, the estimation of culverts and trees will be more accurate if treat the point cloud in separate layers for suspended obstacles problem. All terrain estimation approaches were verified with good performance in an interactive system, which is set up with Virtual Robot Experimentation Platform (VREP) and actual vehicle.

Yang Yi

Papers

Method for identifying objects in 3D point cloud data

Piezoelectric ceramic flapping-wing-type robot

Speed control method of frog board robot

Swinging arm driving-type motion control method for amphibious frog board robot

Real-time terrain estimation based on multi-scale radial basis function for unmanned ground vehicle