Method for humanoid robot to follow accompanying object

Abstract

The invention relates to the technical field of robots, and discloses a method for a humanoid robot to follow an accompanying object, and the method is characterized in that the robot recognizes and tracks the image and position of the accompanying object through a camera or a depth sensor, obtains environment data through a laser radar, an infrared sensor and the like, carries out the path planning of obstacle avoidance through a Dijkstra algorithm, and carries out the tracking of the accompanying object. The speed and direction are adjusted according to object movement to follow in real time, and a proper distance is kept according to steps. The robot integrates vision and environment data to accurately judge the condition, can monitor vital signs of an object, triggers an alarm when finding out abnormity, provides emotional support through voice interaction and the like, gives an alarm, calls rescue and contacts people in emergency, and sends a position when the object leaves a safe area. The system has the advantages that real-time tracking, dynamic path planning, health monitoring and multi-mode interaction of the robot on an accompanying object are achieved through visual identification, sensor fusion and an intelligent algorithm.

Description

Technical Field

[0001] The present invention relates to the field of robotics technology, and in particular to a method for a humanoid robot to follow and accompany a subject. Background Art

[0002] Existing humanoid robots have the following main problems in accompanying scenarios: First, insufficient following ability. Most robots rely on preset paths or semi-automated operations and are unable to track the dynamic movements of the person being accompanied in real time. For example, when the person being accompanied walks freely indoors or changes direction, it is difficult for the robot to quickly adjust its position, resulting in the risk of loss or collision, and inability to provide continuous companionship. Second, poor environmental adaptability. Existing technologies lack real-time perception and decision-making capabilities for complex environments. When encountering obstacles, the robot cannot autonomously plan obstacle avoidance paths and requires human intervention to continue the task, limiting its application in dynamic environments such as homes and hospitals. Third, limited functionality. Traditional accompanying robots only have basic voice interaction or motion control functions, lack the ability to monitor the health status of the person being accompanied, cannot respond to emergencies in a timely manner, and cannot meet the safety care needs of the elderly or disabled.

[0003] Furthermore, existing technologies lack intelligent decision-making mechanisms for multi-sensor fusion, preventing effective coordination between visual and environmental perception data. This results in inaccurate robot judgments of the person being cared for and delayed adjustments to the following strategy. Therefore, developing a humanoid robot with autonomous following, intelligent obstacle avoidance, health monitoring, and dynamic interaction capabilities is a key technical approach to improving the quality of accompanying services. Summary of the Invention

[0004] To solve the above-mentioned problems, the present invention proposes a method for a humanoid robot to follow a caregiver through visual recognition, sensor fusion and intelligent algorithms, which can realize real-time tracking, dynamic path planning, health monitoring and multimodal interaction of the caregiver.

[0005] To solve the above technical problems, the present invention proposes a technical solution: a method for a humanoid robot to follow and accompany a person, comprising the following steps:

[0006] Step 1: The robot uses visual sensors to identify the image of the person being cared for and track their location;

[0007] Step 2: Obtain surrounding environment data through sensors to perform path planning and obstacle avoidance;

[0008] Step 3: Adjust the robot's speed and direction according to the movement of the person being cared for, and follow the person in real time;

[0009] Step 4: When abnormal behavior is detected in the person being cared for, the robot performs emergency response processing.

[0010] Preferably, the visual sensor is a camera or a depth sensor, which can identify and update the position and posture of the accompanying object in real time.

[0011] Preferably, the robot also includes a health monitoring module, which can monitor the vital signs of the accompanying person in real time and trigger an alarm mechanism when an abnormality is found.

[0012] Preferably, the robot interacts with the person being cared for through voice recognition and voice feedback to provide emotional support and life reminders.

[0013] Preferably, the robot adjusts its speed according to the pace of the person being accompanied and always follows at an appropriate distance.

[0014] Preferably, the path planning is implemented using the Dijkstra algorithm using data obtained by laser radar and infrared sensors to ensure that the robot avoids obstacles and selects the optimal path.

[0015] Preferably, the robot fuses the visual recognition data with the environmental data acquired by the sensor to more accurately determine the location of the person being cared for and the surrounding environment.

[0016] Preferably, the emergency response process includes sounding an alarm, calling for rescue, contacting a guardian or medical personnel, and if the robot detects that the person being accompanied has left the safe area, the robot will also send real-time location information.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] Autonomous following capability: Free from human intervention, it can achieve dynamic tracking in complex environments and adapt to various indoor and outdoor scenarios.

[0019] Intelligent environmental adaptation: Multi-sensor data fusion improves path planning accuracy, effectively avoids dynamic obstacles, and reduces collision risks.

[0020] Multi-dimensional safety assurance: Real-time health monitoring and emergency response mechanisms shorten the time to handle abnormal events and ensure the safety of the person being cared for.

[0021] Humanized interactive experience: Multimodal interactions such as voice and touch meet the needs of emotional companionship and life care, and enhance user dependence. DETAILED DESCRIPTION

[0022] Example 1

[0023] The robot uses built-in visual sensors to identify and track the person being cared for. These sensors include a binocular camera and a structured light depth sensor. The camera captures high-resolution images, while the depth sensor acquires three-dimensional spatial coordinate information. In practice, the camera captures the person being cared for in real time, while the depth sensor simultaneously measures the distance and spatial position between the person and the robot. These two sensors work together to accurately determine the person's position and posture. The image data is first preprocessed using Gaussian filtering to remove noise, and then an adaptive threshold segmentation algorithm is used to extract foreground objects and eliminate background interference, ensuring accurate subsequent recognition.

[0024] After image preprocessing, the robot locates the person being cared for using a target recognition algorithm. The system uses the YOLOv5 target detection model, pre-trained with data such as human outlines and clothing features. This model can quickly identify the person being cared for in the image and locate their position within the frame. Simultaneously, the SIFT algorithm extracts key features of the object, such as the coordinates of the shoulders and hips, to create a feature descriptor database for target matching in subsequent frames, enabling continuous tracking of the object. If the object moves or changes posture, the model updates the feature data in real time to ensure continuous tracking.

[0025] During dynamic tracking, the system uses a Kalman filter algorithm to predict the object's trajectory and a MeanShift algorithm to achieve inter-frame target matching, accurately following the movements of the subject. If the subject turns, accelerates, or changes direction, the robot automatically adjusts the camera's gimbal angle to ensure the subject remains centered in the image, maintaining stable tracking. The vision system outputs object position data at a fixed frequency and transmits it to the robot's main control chip via a network protocol, providing real-time data support for subsequent path planning and real-time tracking.

[0026] Example 2

[0027] The robot uses a variety of sensors to acquire data about its surroundings, enabling path planning and obstacle avoidance. These sensors include lidar, infrared obstacle avoidance sensors, and an inertial navigation module. The lidar performs 360-degree scanning, generating point cloud data to construct an environmental map. The infrared sensor detects blind spots in the lidar, and the inertial navigation module records the robot's posture data in real time to ensure coordinate system consistency. The lidar scans the surrounding environment at high frequency. After voxel filtering removes outliers from the point cloud data, a plane fitting algorithm is used to identify obstacles such as tables, chairs, and steps, providing basic environmental data for path planning.

[0028] During the global path planning phase, the robot first uses the SLAM algorithm to construct a global map of the environment. The user then uses the touchscreen to define a safe following area, such as a specific indoor room or a hospital corridor. When the person begins to move, the system uses the Dijkstra algorithm to calculate the optimal path, starting from the robot's current position and ending at the person's real-time coordinates. This algorithm considers factors such as distance and obstacle density to generate a feasible path from the starting point to the end point, ensuring the robot can choose a reasonable route in complex environments.

[0029] During real-time following, if the LiDAR detects an obstacle ahead, such as a moving pedestrian or temporarily stacked items, the robot's local path planning module immediately activates, using the Dijkstra algorithm to generate a detour sub-path based on the global path, achieving dynamic obstacle avoidance. Simultaneously, the robot uses wheel speed encoders to provide real-time speed feedback, and a PID control algorithm to adjust the left and right wheel speeds to ensure smooth steering and avoid collisions. To adapt to the needs of different scenarios, the system dynamically adjusts the following distance using a fuzzy control algorithm: maintaining a safe distance in open spaces and shortening the distance appropriately in confined areas, thereby ensuring timely escort services without affecting the subject's movement.

[0030] Example 3

[0031] The robot integrates multiple physiological parameter sensors to monitor the patient's health and detect abnormalities. These sensors include a wearable heart rate monitoring bracelet, a gait pressure sensor embedded in the shoe sole, and a video surveillance module. The heart rate bracelet uses photoplethysmography technology to monitor heart rate in real time. The gait pressure sensor detects plantar pressure distribution and analyzes walking stability. The video surveillance module uses a skeletal keypoint detection algorithm to identify abnormal movements such as falls. When an abnormal heart rate, unstable gait, or fall is detected, the system immediately triggers an appropriate alarm mechanism, such as a voice reminder, a buzzer alarm, or automatic contact with the guardian, ensuring timely handling of the emergency.

[0032] The robot communicates with the care recipient through multimodal interaction, enhancing the caregiving experience. Voice interaction is based on an end-to-end speech recognition model and supports natural language commands, such as "Take me to the living room" or "Play music." The system can quickly understand and respond to commands. A built-in affective computing module analyzes voice intonation and facial expressions to determine the user's mood and automatically switch interaction strategies, such as playing soothing music or providing reassuring voice commands. Furthermore, the robot is equipped with a touchscreen interface with a large font design, adapted to the operating habits of the elderly. It includes functions such as emergency calls, calendar management, and video calls, allowing users to complete operations with simple clicks, making it quick and convenient.

[0033] When the health monitoring module detects abnormal data, the robot performs an emergency response according to a preset process. For example, if an abnormal heart rate is detected and persists for a certain period of time, the system first reminds the user to remain calm through voice, and simultaneously sends an alert text message containing the real-time location to the preset emergency contact via the network. The video recording function is also activated to record the environmental data before and after the abnormality to provide a reference for subsequent medical analysis. If the accompanying person is detected to have fallen, the system will automatically call the emergency number to ensure that professional assistance is received in the shortest possible time. The entire abnormal response process is efficient and orderly, and the time from detecting the abnormality to completing the alarm transmission is controlled within a relatively short range, which significantly improves the efficiency of emergency response and ensures the safety of the accompanying person.

[0034] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for a humanoid robot to follow and accompany a person, characterized in that: The following steps are involved: Step 1: The robot uses visual sensors to identify the image of the person being cared for and track their location; Step 2: Obtain surrounding environment data through sensors to perform path planning and obstacle avoidance; Step 3: Adjust the robot's speed and direction according to the movement of the person being cared for, and follow the person in real time; Step 4: When abnormal behavior is detected in the person being cared for, the robot performs emergency response processing.

2. The method for a humanoid robot to follow and accompany a person according to claim 1, characterized in that: The visual sensor is a camera or a depth sensor, and the sensor can identify and update the position and posture of the accompanying object in real time.

3. The method for a humanoid robot to follow and accompany a person according to claim 1, characterized in that: The robot also includes a health monitoring module that can monitor the vital signs of the person being cared for in real time and trigger an alarm mechanism when an abnormality is detected.

4. The method for a humanoid robot to follow and accompany a person according to claim 1, characterized in that: The robot interacts with the person being cared for through voice recognition and voice feedback, providing emotional support and life reminders.

5. The method for a humanoid robot to follow and accompany a person according to claim 1, characterized in that: The robot adjusts its speed according to the pace of the person being escorted and always follows at an appropriate distance.

6. The method for a humanoid robot to follow and accompany a person according to claim 1, characterized in that: The path planning is implemented using the Dijkstra algorithm using data obtained from lidar and infrared sensors to ensure that the robot avoids obstacles and chooses the optimal path.

7. The method for a humanoid robot to follow and accompany a person according to claim 1, characterized in that: The robot integrates the visual recognition data with the environmental data acquired by the sensor to more accurately determine the location of the person being cared for and the surrounding environment.

8. The method for a humanoid robot to follow and accompany a person according to claim 3, characterized in that: The emergency response process includes sounding an alarm, calling for rescue, and contacting guardians or medical personnel. If the robot detects that the person being cared for has left the safe area, it will also send real-time location information.