See the View (STV) Project Environmental Perception for Safer and - PowerPoint PPT Presentation

See the View (STV) Project Environmental Perception for Safer and Reliable Autonomous Driving Arie Gavriely Arie.Gavriely@vayavision.com Co-funded by the Horizon 2020 program of the European Union Proprietary & Confidential | 1 Proprietary & Confidential | 1

What we do Vehicle Perception based on raw data fusion with upsampling and AI Driving Decisions Perception & Fusion Sensors Co-funded by the Horizon 2020 program of the European Union

Fusion & Perception ⁺ Raw data fusion not objects Camera Perception Engine ⁺ 3D RGBd model LiDAR Raw Data Fusion & ➢ Higher detection rates Detection RADAR ➢ Less false alarms ➢ Overcome faults . . . Co-funded by the Horizon 2020 Proprietary & Confidential | 3 program of the European Union

Sensors Raw Data ⁺ Camera ⁺ Lidar ⁺ Radar Fusion & Detection Framework ⁺ Higher Quality (accuracy and reliability) ⁺ Lower system cost (Sensors, BUS, processing, energy) Environmental Model Co-funded by the Horizon 2020 Proprietary & Confidential | 4 program of the European Union

Fusion & Perception Camera Objects data-base Road Detection Obstacle detection HD RGBd Calibrate Object BB detection 3D Model Up- Objects LiDAR sampling Semantic Segment. & E-M Unroll I/F Ego- & Multi-frame Object motion Tracking Driving Environmental Model Match RADAR Module Raw data Fusion Detection, Classification, Segmentation . . . Software Framework Co-funded by the Horizon 2020 Proprietary & Confidential | 5 program of the European Union

LIDAR 64 - Before Up-sampling Co-funded by the Horizon 2020 program of the European Union

LIDAR 64 - After Up-sampling 3D High Resolution at Low Cost Co-funded by the Horizon 2020 program of the European Union

Accurate SLAM – Ego-Motion Co-funded by the Horizon 2020 Proprietary & Confidential | 8 program of the European Union

Environmental Model Detection by Non- DNN Bird Eye View Red – Objects Blue Ground Free Space Vehicles Detection by Classification (DNN) Free Space Pedestrian Co-funded by the Horizon 2020 Proprietary & Confidential | 10 program of the European Union

Peacocks crossing the street Detecting the unexpected – Non-DNN Proprietary & Confidential | 11

Urban Driving in Tel-Aviv Proprietary & Confidential | 12

STV Project Objectives Co-funded by the Horizon 2020 Proprietary & Confidential | 13 program of the European Union

STV expected and planned performances Co-funded by the Horizon 2020 Proprietary & Confidential | 14 program of the European Union

STV Project plan Co-funded by the Horizon 2020 Proprietary & Confidential | 15 program of the European Union

STV Schedule and Gantt Chart Co-funded by the Horizon 2020 Proprietary & Confidential | 16 program of the European Union

⁺ Raw Data Fusion ⁺ The optimal method for reliable detection ⁺ Up-Sampling ⁺ Increasing LiDAR/RADAR low resolution with smart algorithms ⁺ Detection and Classification Benefits ⁺ Most accurate and reliable detection Better performance with current sensor set and Roadmap for low cost system Co-funded by the Horizon 2020 Proprietary & Confidential | 17 Proprietary & Confidential | 17 program of the European Union