CN-121977534-A - Mobile robot navigation hybrid map construction method, device and medium

CN121977534ACN 121977534 ACN121977534 ACN 121977534ACN-121977534-A

Abstract

The invention discloses a method, equipment and medium for constructing a hybrid map for mobile robot navigation, which relate to the technical field of navigation hybrid map construction and comprise the steps of collecting multi-mode sensor observation flows, constructing metric characterization, topological characterization and semantic characterization, generating a time version chain starting point time version, updating the metric characterization, the topological characterization and the semantic characterization in a control period, generating differential representation, writing the differential representation into a time version chain, generating a continuously-expanded time version chain, generating structural stability, semantic drift degree, topological evolution consistency and updating cost density quality curves, acquiring a quality curve set, selecting a current running hybrid map time version, carrying out local updating of the metric characterization, the semantic characterization and the topological characterization, and adjusting a time window length and a time version switching strategy. The objectivity of map quality evaluation and time version selection decision precision are improved, and the map updating efficiency and the overall operation efficiency are improved.

Inventors

WU JIAN
MA YE
WANG FEI
YUAN MENGRU
SHI MIAO

Assignees

杭州它人机器人技术有限公司

Dates

Publication Date: 20260505
Application Date: 20260403

Claims (10)

1. A method for constructing a hybrid map for mobile robot navigation is characterized by comprising the steps of, Collecting multi-mode sensor observation flows, performing time alignment and coordinate unification, generating a fusion observation frame, constructing measurement characterization, topology characterization and semantic characterization, combining to generate a hybrid map initial version, and generating a time version chain starting point time version; Based on the starting time version of the time version chain, updating metric characterization, topology characterization and semantic characterization in a control period, recording update events, generating differential representation, writing the differential representation into the time version chain, and generating a continuously-expanded time version chain; generating a quality curve of structural stability, semantic drift degree, topological evolution consistency and update cost density according to the continuously-expanded time version chain, and obtaining a quality curve set; And selecting the current running hybrid map time version according to the quality curve set, carrying out local updating of measurement characterization, semantic characterization and topology characterization, and adjusting the time window length and the time version switching strategy.
2. The method for constructing a hybrid map for mobile robot navigation according to claim 1, wherein the steps of collecting the multi-mode sensor observation stream, performing time alignment and coordinate unification to generate a fusion observation frame, Collecting space geometrical observation data, environment semantic observation data and odometer observation data, and generating a multi-mode sensor observation stream; Constructing a reference time axis, sequentially performing monotonicity check, frame loss detection, frequency alignment and clock drift correction on the multi-modal sensor observation stream, and generating a time-aligned multi-modal sensor observation stream; and performing coordinate conversion on the time-aligned multi-mode sensor observation stream to obtain the time-aligned and coordinate-unified multi-mode sensor observation stream, and generating a fusion observation frame.
3. The method for constructing a hybrid map for mobile robot navigation according to claim 2, wherein the constructing the metric representation, the topology representation and the semantic representation, combining to generate a hybrid map initial version, generating a temporal version chain start temporal version refers to constructing the metric representation, the topology representation and the semantic representation by occupying a grid according to a fusion observation frame, combining to generate a hybrid map initial version, establishing a cross-representation alignment index, and writing the temporal version chain start temporal version.
4. The method for constructing a hybrid map for mobile robot navigation according to claim 3, wherein the updating of metric characterization, topology characterization and semantic characterization in a control period based on a temporal version chain starting point temporal version, recording update events comprises the steps of, The method comprises the steps of obtaining a fusion observation frame of a current control period, updating a measurement representation in an occupied grid index space according to the sequence of addition, deletion and correction, obtaining a measurement representation change event sequence, and generating a measurement representation change occupied grid index set; deriving environment communication observation in the measurement representation change occupation grid index set, and updating topology representation according to the sequence of node updating, edge updating and consistency checking to obtain a topology representation change event sequence; and updating the semantic representation in the grid index set according to the sequence of area update, entity update and category stable rechecking in the measurement representation change occupation grid index set to obtain a semantic representation change event sequence.
5. The method for building a hybrid map for mobile robot navigation according to claim 4, wherein the step of generating the differential representation, writing the differential representation into the time version chain, generating the continuously extended time version chain comprises the steps of, Generating a differential representation based on the metric characterization change event sequence, the topology characterization change event sequence, and the semantic characterization change event sequence in combination with the cross-token alignment index change event sequence; and according to the differential representation, expanding the time version chain, binding the time window boundary, generating an archive segment, and acquiring the continuously expanded time version chain.
6. The method for constructing a hybrid map for mobile robot navigation according to claim 5, wherein the steps of generating a quality curve set of structural stability, semantic drift, consistency of topological evolution and update cost density according to a continuously extended time version chain are as follows, Constructing a time window event sequence, respectively calculating and normalizing a structural stability basic value and a semantic drift basic value, and generating a structural stability quality curve and a semantic drift quality curve; Based on the time window event sequence, calculating and normalizing a topological evolution consistency basic value and an updating cost density basic value, generating a topological evolution consistency quality curve and an updating cost density quality curve, and obtaining a quality curve set.
7. The method for constructing a hybrid map for mobile robot navigation according to claim 6, wherein said selecting a current running hybrid map time version based on a set of quality curves comprises the steps of, Acquiring a current control period time stamp, determining a target time window number and a time version number interval, reading a quality curve set, and generating a historical sequence of structural stability, semantic drift degree, topological evolution consistency and update cost density; and calculating a history reference value based on the structure stability, the semantic drift degree and the history sequence of the topological evolution consistency degree, and selecting the current running hybrid map time version.
8. The method for constructing a hybrid map for mobile robot navigation according to claim 7, wherein said local updating of metric characterization, semantic characterization and topology characterization adjusts a time window length and a time version switching strategy by the steps of, Triggering the local update of the metric characterization, the semantic characterization or the topology characterization according to the current running hybrid map time version; And adjusting the time window length configuration parameters and the time version switching strategy according to the updated cost density history sequence.
9. A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the method is characterized in that the processor realizes the steps of the hybrid map construction method for mobile robot navigation according to any one of claims 1-8 when executing the computer program.
10. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor realizes the steps of the hybrid map construction method for mobile robot navigation according to any one of claims 1 to 8.

Description

Mobile robot navigation hybrid map construction method, device and medium Technical Field The invention relates to the technical field of navigation hybrid map construction, in particular to a mobile robot navigation hybrid map construction method, equipment and medium. Background The mobile robot relies on unified modeling of the space structure, the communication relation and the semantic attribute of the surrounding environment to realize autonomous navigation, a multi-sensor fusion technology is adopted, multi-source observation data such as laser, vision and odometer are integrated, a hybrid map comprising occupied grids, topological nodes and semantic tags is constructed, the hybrid map provides critical information such as passable areas, path connectivity and environmental objects for the mobile robot, and the mobile robot is supported to position, plan paths and make decisions in a structured or semi-structured scene. In the operation scene of multi-source observation, the conventional method for constructing the hybrid map still has room for improvement in the time evolution dimension, on one hand, the conventional method for constructing the hybrid map focuses on local increment updating and consistency correction, the quantitative expression of the structural stability, semantic consistency and topological evolution consistency of a cross-time window is insufficient, and the quality of the map is difficult to be evaluated in a system in long-term evolution, and on the other hand, the updating strategy adopts a fixed period mechanism, the time version switching and evolution chain management are relatively static, and a dynamic decision mechanism is lacked, so that the self-adaptability of the navigation system under a complex observation sequence is limited. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a mobile robot navigation hybrid map construction method which solves the problems of insufficient time version chain evolution expression system and lack of cross-time window quality quantification and regulation capability in the prior art. In order to solve the technical problems, the invention provides the following technical scheme: in a first aspect, the present invention provides a hybrid map construction method for mobile robot navigation, comprising, Collecting multi-mode sensor observation flows, performing time alignment and coordinate unification, generating a fusion observation frame, constructing measurement characterization, topology characterization and semantic characterization, combining to generate a hybrid map initial version, and generating a time version chain starting point time version; Based on the starting time version of the time version chain, updating metric characterization, topology characterization and semantic characterization in a control period, recording update events, generating differential representation, writing the differential representation into the time version chain, and generating a continuously-expanded time version chain; generating a quality curve of structural stability, semantic drift degree, topological evolution consistency and update cost density according to the continuously-expanded time version chain, and obtaining a quality curve set; And selecting the current running hybrid map time version according to the quality curve set, carrying out local updating of measurement characterization, semantic characterization and topology characterization, and adjusting the time window length and the time version switching strategy. The invention relates to a method for constructing a hybrid map for mobile robot navigation, which comprises the steps of collecting multi-mode sensor observation flows, performing time alignment and coordinate unification to generate a fusion observation frame, Collecting space geometrical observation data, environment semantic observation data and odometer observation data, and generating a multi-mode sensor observation stream; Constructing a reference time axis, sequentially performing monotonicity check, frame loss detection, frequency alignment and clock drift correction on the multi-modal sensor observation stream, and generating a time-aligned multi-modal sensor observation stream; and performing coordinate conversion on the time-aligned multi-mode sensor observation stream to obtain the time-aligned and coordinate-unified multi-mode sensor observation stream, and generating a fusion observation frame. The method for constructing the hybrid map for the mobile robot navigation is characterized in that the construction metric characterization, the topology characterization and the semantic characterization are combined to generate a hybrid map initial version, the generation of a time version chain starting point time version refers to the construction of the metric characterization, the topology character