images

AVP Control – Outdoor Driving Technology

CubeAi has ACC, AEB, Lane Change, and AVP technology.

Adoptive Cruise Control (ACC)

Section : Straight and Curved section of roads with both lanes
State : Automatically adjusts the vehicle speed to maintain a safe distance from vehicles ahead. Comfortable and relaxed driving, even in heavy traffic jam
Sensor : Camera , GPS, Radar, LiDar

Automatic Emergency Braking (AEB)

Section : Straight and Curved section of roads with both lanes
Target Speed : Speed under 100 kph
Sensor : GPS, Radar, LiDar

Automatic Lane Change / Blind Spot Detection

Section : Straight and Curved section of roads with both lanes
State : Automatically take the correct freeway exit or interchange based on the route entered into the navigation system. Also, detects other vehicles located to the driver’s side and rear.
Sensor : GPS, Radar

images

AVP Control – Indoor Driving Technology

Light Guidance System

CUBE AI provides Autonomous Valet Parking service at indoor parking lot without GPS. This causes Autonomous Driving Era One step ear

Intelligence Infrastructure as A Guidance for all existing cars In present

CUBE AI has been developed intelligent parking lot infrastructure system for reducing parking stress which guides the car to right-spot to park using CUBE AI’s service. This is one of CUBE AI’s Short-Term implementation model to reach new customers segment.

This is New complementary service , means that the development team has unique composition of idea for mobility service even now. Aim to being a Leader of mobility service is one of CUBE AI philosophy.

LED Light guides the cars from entrance of parking lot to empty parking space in Big mart or shopping mall. No parking attendances are needed. CUBE AI’s Light-Guidance System guides the cars with algorithm technology.

AVP Control – Indoor Driving Technology

LiDar is applied on CUBE AI’s AVP system currently.
However, AVP system of CUBE AI’s will be implemented with only ‘Vision AI Technology’ in AVM.

Required Sensor systems for AVP

GNSS / INS
Detect the current position of vehicle in real-time
Not Required for CUBE AI
RADAR (ESS)
Detect obstacles in front & rear of vehicle. Long range recognition
Not Required for CUBE AI
LiDAR(LDR-16)
Sensors to detect obstacles and pedestrians around the vehicle, front, rear and left and right covers
Currently Applied for CUBE AI
CAMERA (Vision Camera)
Intuitive recognition of forward information
Not Required for CUBE AI
AVM (Around View Monitor)
Detect parking-guidance line and parking lane images
Developing ‘Vision AI Technology’ The final goal of Cube AI : AVP system with VISION AI Technology

Indoor AVP control system requires APM for Lateral control by signaling to MDPS, ASM for control accelerator by signaling to accelerator position sensor, ASM for control speed-reduction by signaling to brake actuator, AGM for Transmission control by signaling on TCU,VCU for Control of integration module for APM,ASM and AGM and Controller (EAS) for detect the current position of vehicle in real-time.

images

Indoor AVP Sequence Action

Step 1. Searching Parking space

Parking lot Map Building

Generate Parking lot map for AVP
Assignment ID by individual parking zone
Control of LED light through identification of parking zone ID

Step 2. Generating for Line detection & Steering angle

AVM(Around View Monitoring) LED Line Tracking

Classification of LED color through AVM image process
(ex. green : run , red : stop)
Detection of driving color LED line and speed and steering angle
Move to allocated parking area with LED line tracking
Generating car location initialization after arrival of allocated parking zone

Step 3. Dead Reckoning & Motion Planning

Dead Reckoning & Motion Planning

Transfer control issue to Motion tracker after initialization of vehicle position
Conform of Local Position through massage initialization
Generate forward & reverse travel routes
Path Tracking using Dead Reckoning
Generate of Order for Stop & Change speed
Transfer control issue to Line tracker after confirming of rear path tracking and entering parking area

Step 4. Detection of lane of Parking area

End Point Parking

Confirming of transfer control issue to Line Tracker
Parking line detection with rear AVM image processing
Creation of Stop order/ Switching gear position and steering angle
Transfer completing of parking signal

images

AI Technology

Recognizing objects and people inside parking lot and planning safest path to parking space to prevent accident. Also, the technology provides AVP training & Vehicle positioning in Non-GPS environment.

3D Space Recognition & Tracking
Driving Vehicle can recognize the road inside parking lot through Spatial Temporal Knowledge

Automated Vehicle Control based on Deep Neural Network
To control self-driving car via modelling of Deep Neural Network

Parking Motion Planning
Vehicle can do ‘Motion Planning’ to park itself to empty space in an effective way and keeps learning the way by itself

AI Puzzling Net
Most of the parking lots consist of the below 8 pieces of puzzle
Cube AI enables vehicle park itself by all combinations of the below puzzles through Deep Neural Network. (Patent applied)

Seamless Tracing Net

Even though the recognized object disappears for a moment, vehicle can continuously trace it by calculating its projected course with the technology of Cube AI

V-Masking Technology

Vehicle does not need to have GPU process. V-Masking of Cube AI enables vehicle to recognize parking line, road in parking lot at high speed.

images images

Control & Network Technology

The team is developing control algorithm to park several vehicles safely all at once and making path after analyzing surrounding information where people and objects coexist as well. CUBE AI delivers V2X, I2I network technologies.

Architecture
Real-Time Object Detection using Cloud platform on product level

Real-Time Object Recognition

Real-Time Object Detection Architecture

Indoor Positioning Technology with Node Edge Map

CUBE Path planning Algorithm development team puts together management and control system technologies. Team needs to include sensor technology with IT technology to calculate the driving maneuver as well as cloud connection to communicate with cars.

Securing of Indoor Positioning technology with GPS or without GPS. The needed technologies such as sensors, dead reckoning, camera, telecommunicate are obtained by CUBE AI based positioning, dead reckoning, Map Matching technology.

Different type of map is needed for underground parking lot The current detailed map for self-driving system costs great deal of time and expenses. CUBE AI implements automated parking with Node Edge Map easily made.

CUBE AI’s digital mapping technology called Node-Edge Map brings the complete map implementation within three hours (100M* 100M)

AI Camera Technology

Competitors’ Camera - Specification