4results about How to "Self-directed learning" patented technology

The present application relates to the technical field of criminal case data analysis and early warning, in particular to a multi-technology coordinated criminal case intelligent early warning system and method fusing pre-training large model customization adaptation, dynamic weighted clustering and semantic vector retrieval, belonging to the cross technical field of judicial big data and artificial intelligence, through the coordinated technical path of "large model label extraction-weighted clustering-vector deduplication", a full-process intelligent system of "data input-intelligent analysis-accurate early warning-feedback iteration" applicable to judicial case analysis and processing scenes is constructed, not only effectively solving the core problems of low efficiency and early warning redundancy under the traditional manual analysis mode, but also completely solving the many common problems existing in the current technical system of criminal case analysis and early warning.

The application relates to a classification sampling tool and method suitable for a low-visibility seabed environment, which comprises a sampler for collecting samples, the structure of the sampler comprising a sampling cylinder, one end of the sampling cylinder being closed by a sealing plate, and a matching rod being connected to the sealing plate; a sample barrel provided with a barrel body for containing the collecting sampler; a cover capable of moving along the axial direction of the sample barrel and controllably closing the barrel body; a limiting structure arranged on the sample barrel and capable of controllably limiting the axial movement of the cover on the sample barrel; and a matching hole corresponding to the matching rod being arranged on the cover. Through the above operation, in-situ classification sampling in a low-visibility environment can be completed, the sound and light pollution of the sampling process to the seabed environment is effectively reduced, the cross contamination between different types of sampling samples is effectively reduced, the classification sampling of samples with different requirements for sample treatment is facilitated, the structure is simple, the sampling operation process is simplified, intelligent sampling operation is facilitated, and the application has strong practicability.

The invention relates to the technical field of robot intelligent control and machine vision, and discloses a robot binocular vision motion mapping method and system, and the method comprises the steps: generating scene perception data containing the three-dimensional information of a target object and joint points of two arms of a human body based on the obtained binocular vision input information and a current operation task instruction; extracting space-time motion features of joint points of two arms of the human body from the scene perception data; and the task-related visual features in the scene perception data are fused with the space-time motion features to obtain fused features, and the fused features are mapped into corresponding robot joint motion tracks through a sequence mapping model. Therefore, the robot motion mapping accuracy and flexibility can be improved.

The application discloses a kind of based on embodiment intelligent large model driving's inspection autonomous learning method, device, equipment and medium, it is related to artificial intelligence technical field, method includes: control multi-sensor acquisition current inspection scene original data and extract feature data, original data includes scene data and / or equipment state data;Feature data and the priori knowledge matching of energy inspection association of light model;When similarity is less than preset threshold, trigger autonomous learning, generate new inspection operation rule for unknown scene, or generate new fault judgment basis for unknown fault;Based on matching result or autonomous learning result, output action instruction;Inspection robot executes instruction and collects entity interaction feedback data;Calibrate action according to feedback data, simultaneously feedback data, new rule or new criterion are uploaded to cloud, and incremental update is carried out to large model, and learning closed loop is formed.The application improves the flexibility and intelligent level of inspection.