Techonology_杭州宏景智驾科技有限公司

Home

Solution

Techonology

About Us

Join Us

Zh En

Zh En

Home

Solution

Product Center

IPM

HDC

T-BOX

Feature Demo

Techonology

About Us

Join Us
Techonology
Techonology
Cloud Platform Simulation Testing Perception Map Localization
Cloud Platform
Data Acquisition
Data Acquisition

Cloud-based vehicle data collection task deployment

Data Acquisition
Data Processing
Data Processing

Extracting data slices based on diverse requirements post-collection

Data Processing
Data Annotation
Data Annotation

Online dataset filtering and curation

Data Annotation
Model Training
Model Training

Cloud-based algorithm model training

Model Training
Data Management
Data Management

Multi‑mode data management:Perception algorithm acquisition data
Long‑distance test mileage data、Test database (DB) data

Data Management

Data Acquisition

Data Processing

Data Annotation

Model Training

Data Management

Simulation & Test
SIL – Software-in-the-Loop
SIL – Software-in-the-Loop
Covers 80% of ADAS longitudinal/lateral control and warning scenarios
Cloud batch automation
HIL (Hardware-in-the-Loop)
HIL (Hardware-in-the-Loop)
Covers 80% of ADAS longitudinal/lateral control and warning scenarios
Automated log analysis
Logsim (Data Replay)
Logsim (Data Replay)
100,000 km/week regression testing
Automated Replay Analysis
95%+ replay accuracy
图片
Perception

Front View

Surround View

Tri-Net Fusion BEV Perception Algorithm

Dynamic Objects
Maximum stable perception distance: 150m
Position accuracy: 5% error within 100m longitudinally
Supports ACC, FCW, AEB and more
Lane Lines
Maximum stable perception distance: 120m
Lateral accuracy: <10cm within 50m
Supports LDW, LKA, LCA and more
Front View
Front View
Dynamic Obstacles
Maximum detection range: 20m
Position accuracy:
Vehicle average error: <8%
Pedestrian average error: <5%
Static Obstacles
Maximum detection range: 12m
Position accuracy:
Within 2m: 5–10cm
2–5m: <5%
Beyond 5m: <10%
Parking Slot Corners
Accuracy when camera and slot are on the same plane:
Within2.5m: up to5 cm
Within 5 m: up to 10 cm
Surround View
Surround View
Sensor Platformization
Supports various sensor inputs including front view, rear view, surround view, fisheye, LiDAR, 4D Radar, and more.
Data Closed-Loop System
A closed-loop ground truth system based on Hyper-GTMax, supporting scenario mining, auto-labeling, and perception KPI evaluation.
Perception Task Platformization
Unified modular output supporting object detection, map detection, occupancy, HD-map-free solutions, and general obstacle detection capabilities.
Software & Algorithm Platformization
Proprietary deployment pipeline and toolchain enabling fast adaptation to Ambarella CV3/CV72, NVIDIA Orin, Horizon J5, and other mainstream platforms
Tri-Net Fusion BEV Perception Algorithm
Tri-Net Fusion BEV Perception Algorithm
Map Localization
Mapping phase: Creates map using sensor and positioning data
Generates reference route for trajectory tracking
Real-time localization matching with memorized route upon parking start
Cruising phase: Retrieves pre-built map and reference line from backend
Provides continuous real-time positioning information
Map Localization
Leading the way
to Autonomous Driving
Join Us
This website uses cookies
This website uses cookies to improve your experience while you navigate through the website. You can choose to opt-out if you wish. For more information, please refer to our 《Cookies》。