Core Tech_浙江通通达科技有限公司

Home

Application

Delivery Bots Sanitation Bots Charging Bots

Core Tech

Operation Services

Service Method Service Content Remote Platform Mobile APP Terminal

Join Us

Contact

中文

Home

Application

Delivery Bots

Sanitation Bots

Charging Bots

Core Tech

Operation Services

Service Method

Service Content

Remote Platform

Mobile APP Terminal

Join Us

Contact

Core Tech

Multimodal Space-Time Sensing Fusion System

The system integrates multi-source heterogeneous data and has multimodal perception capabilities. By means of a front-fusion model and combined with a high-quality dedicated dataset for unmanned delivery vehicles independently constructed, it realizes accurate target detection, dynamic tracking and trajectory prediction. In various typical application scenarios, the system demonstrates all-weather stability and high robustness in complex environments, and can effectively cope with typical unmanned delivery scenarios such as urban non-motorized lanes, intersections, low-light conditions at night, and rainy/foggy weather.

Data Collection and Simulation Closed-Loop Platform

The environmental information data collection system based on the AI large-model perception model can complete functions such as data synchronization module and standard-format perception dataset generation while performing intelligent driving. Online, it is equipped with functions including automatic annotation module and 4D annotation module, supporting long-term data asset construction and data mining.

High-Precision Mapping and Multi-Source Positioning System

The multi-source data mapping and positioning system matches with offline maps multiple times from coarse to fine to continuously correct radial errors in real time. Through continuous matching and calibration with offline maps, it corrects angular errors promptly when they occur, reducing the lateral length error caused by angular errors to less than 1 cm and ensuring the coincidence of round-trip routes. The mapping and positioning accuracy of the platform is less than 5 cm.

Logistics Large Model Intelligent Operation Platform

The Tongtongda Operation Platform connects to a large number of unmanned devices and adopts an intelligent hub based on the logistics large language model, significantly improving the efficiency of delivery scheduling.

Operational Services

Supervision Box
Fleet Platform
Station Management
Monitoring & Operation
Cloud Driving Operation Service

Intelligent Big Data

Data Lake Management and Analysis Platform EDAP
ElasticsearchBES

Intelligent Driving Data Management

Collection Scheduling Management
Automatic Data Upload
Source Data Management
Dataset Management
Workflow Engine
IF Data Visualization

Foundation Model

Underwent continuous logistics-specific pre-training based on DeepSeek-67B, with a professional knowledge graph (2 million logistics entity relationships) integrated.
Intelligent Customer Service: Addresses over 70% of customer inquiries and accurately parses unstructured descriptions (e.g., "place in the 3rd layer of the property locker").
Continuous Learning: Features a feedback reward model for daily incremental updates to delivery strategy knowledge.

Operational Improvement

Scheduling efficiency increased by 35%, customer complaint rate decreased by 60%, and manual intervention reduced by 80%.

Digital Twin

A hierarchical curriculum learning framework is constructed. The basic layer learns regular delivery in a digital twin environment, and the PPO-λ algorithm is used to achieve smooth policy transfer. A policy feature library with a cross-city topological structure is established, requiring only 10% of real data for fine-tuning during new scenario deployment.

Training Efficiency

The adaptation cycle for new urban scenarios has been shortened from 3 months to 2 weeks, and the demand for training data has decreased by 80%.

Build an AI+embodied intelligence terminal platform<br>Transform human lifestyles

Build an AI+embodied intelligence terminal platform
Transform human lifestyles

Consult about products