Driving with AI: The Future of Mobility and Transportation Harsha - PowerPoint PPT Presentation

Driving with AI: The Future of Mobility and Transportation Harsha Badarinarayan , Ph.D. Director and Laboratory Manager Global Center for Social Innovation – North America, Hitachi America, Ltd.

Overview § Hitachi – Leaders in Digital Innovation § Mobility Use Cases i. Connected Powertrain ii. Ride Comfort Control using AI iii. Predictive Maintenance iv. Vehicle Occupant Health and Safety Analytics § Inspiring the Next…..

Hitachi – a digital innovation company As a global technology leader, digital innovation is at the heart of everything we do. Hitachi is investing $2.8B over 3 years to build a global leadership in IoT Heritage of innovation $88.8B $3.5B 15,000+ 119,000 global customers Industrial consolidated revenue* annual R&D* and IT Leader global patents 1,400+ #79 11,000 $5.4B $2.8B strategic alliances IoT revenue* on Fortune 500 IoT patents 3-year IoT R&D* Thomson Reuters Award 110+ years 1,000+ Top 100 winning technology and subsidiaries 15,000 Global Innovators product development culture global partners 335,000 employees * As of March 31, 2016

The Era of Digital Disruption Presenting existential threats and transformative opportunities for Fortune 500 Global Mega Trends Globalisation Aging Populations Mass Migration Urbanisation Sustainability Terrorism Cyber Crime Automation IoT Connected AI Peer-to-peer Social Technical 3D Printing Automated Innovation Innovation Blockchain As-a-Service Cloud Shared Big Video Creating new markets; breaking down industry boundaries; and disrupting value chains Society 5.0 Manufacturing Public Energy Automotive Financial Rail and Industry Service Service

Major Trends Disrupting the Automotive Industry Non-traditional Light-weighting Connectivity & AI Shift to Asia entrants Combustion engine Electrification Alternative Fuels Digital Experience advancements Short-term trends Mid-term trends Big Data/ Shared Mobility Low cost brands Long-term trends Analytics New Retail – Direct Start-up OEMs to Customer Intelligent and Fuel Cells Automated Vehicles Source: Roland Berger

Future of Mobility… Source: Roland Berger

i. Connected Powertrain

Fuel Economy Improvement with Connected Powertrain Realize eco-friendly propulsion system by creating synergy between powertrain technology AD/ADAS and Connectivity for safety and comfort AD/ADAS Features Powertrain Control E.g. Start & Stop control ACC/AEB (Preceding vehicle recognition) Opportunity for Start & Stop Sailing stop Lane keep system (lane recognition) Coasting stop Autonomous Driving for highway × Idling stop VSP AD/ADAS Powertrain AD: Autonomous Driving Technologies Control Vehicle Speed ADAS: Advanced Driver Assistance System Driving Cycle (Acceleration suppression) ACC: Adaptive Cruise Control HEV control (Regeneration) AEB: Autonomous Emergency Braking Synergy to create eco-friendly propulsion systems

Connected Powertrain Architecture Connected Connected Car as a Vehicle Controls Sensor Road Grade, Drive Cycle Engine Map HD LDM + V2X + Traffic Weather, Traffic Estimating accurate position via HD Map Information Weather- robust Vision System and Sensor Fusion Architecture RADAR/LiDAR Real World Driving

� � AI Neural Network Can Reduce Average Fuel Consumption by 10% Optimize Training Drive Cycle Outcomes Neural Net Supervisory Control Estimation Predicted Drive Cycle Vehicle Speed (mph) Input layer Output layer § Emission Feature Extraction reduction - Trip Time § CAFÉ Improvements - Avg. Speed Extract § Gas Savings Features - Congestion Level + § Comfort Ride - Road Type Live Traffic Time (Secs) min (𝐺 𝑦, 𝑣 = 𝑛𝑗𝑜 . 𝑔𝑣𝑓𝑚_𝑠𝑏𝑢𝑓(𝑄 789 , ω 789 |𝑄 <=>?7=_=7@ )

Traffic & Connected Powertrain Co-simulation § Perform large scale traffic simulation using Real World Driving data § Evaluate benefit of ‘look-ahead’ controls & calibration for powertrain Multiple co-simulation platform Large-scale traffic on Michigan road network Speed (mph) Free Flow Traffic Speed Throttle Control Brake Control Predicted Traffic Speed (mph) 3D vehicle dynamics § Traffic Flow § Vehicle Velocity

ii. Ride Comfort Control using AI

Value of an Autonomous car : Safety + Experience + Comfort Safety Drive Experience Ride Comfort Ex: Driver Assist Technology Ex: Configurable seats Ex: Semi-Active Dampers Our AI based approach for Enhanced Ride Comfort Safe, Smooth & Comfortable Ride Maps/Traffic Road Topology AI – Machine Learning Sensing Technology Fusion coupled with AI Road Profile Dynamic Localization

Ride Comfort Solution: Pitching – Acceleration/Deceleration Yaw / Roll – Turns & Curved Roads Image: DSPORT Magazine

Development of AI Platform

System Architecture – Turn Prediction Input Approach Output Identify GPS Feature Extraction Model Training/Test Driver Calculate Drive OBD Speed G-Force Mode Polish Trained Predict Accel/ Gyro_Z Features Model Turn Gyro GPS Heading

Road Tests – Drive Model & Turn Prediction

iii. Predictive Maintenance

Need for Predictive Maintenance – Background Proactive and predictive maintenance approach are needed to minimize warranty and service related expense for next generation mobility: ‒ Reactive maintenance approach no longer relevant/applicable in this landscape ‒ Need Proactive and Predictive maintenance ‒ Use-case of a shock absorber to emphasize the predictive maintenance benefits for reducing warranty costs Consumer/Commercial Vehicles Fleet Vehicles Warranty Costs as a Percentage of Annual Sales Source: ELEMENT Source: WarrantyWeek

Predictive Maintenance Architecture IoT Architecture with Edge Analytics to Collect, Process, Analyze & enable Decision Making Capability through Cloud Infrastructure

Results Visualization – Miles to Failure Prediction Potential Reduction in Component Life

iv. Vehicle Occupant Health and Safety Analytics

Overview – Vehicle Occupant Health and Safety Background and Value v Market projections $7.74B by 2020 Requirements and Challenges Before Driving – Breath Alcohol Detection v Effective distraction monitoring v Remote health monitoring v Estimate driver fatigue level Our Key Differentiation ü Breath Alcohol Detection ü Facial state recognition ü Fusion of physiological sensors During Driving – Monitor Psychophysiological State and vision system ü On-going field trials in the US

Breath Alcohol Detection System ü Multi-gas Sensor ü Facial Recognition ü 3X More Accurate ü Abuse Prevention Drunk Driving Accounts for 30% of Features – Anywhere, Anytime, and Easy Measurement . Humidity and Total Deaths in USA Multi-gas Sensor distinguish human breathe from artificial gas Data log sheet 6月1日 8: 30 2. 45 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月2日 8: 31 2. 14 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月3日 8: 32 1. 58 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月4日 8: 33 2. 16 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月5日 8: 34 2. 45 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月6日 8: 35 2. 14 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月7日 8: 36 1. 58 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月8日 8: 37 2. 16 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月9日 8: 38 2. 45 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月10日 8: 39 2. 14 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月11日 8: 40 1. 58 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ A Driver Management System Could 6月12日 8: 41 2. 16 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月13日 8: 42 2. 16 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ 6月14日 8: 43 2. 16 ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ○ Improve Business Efficiency Especially for Transport Industry Utilizing Breathe Alcohol Detection System

Occupant Psychophysiological State Monitoring In-vehicle sensor fusion Cloud Platform Real-Time Data Usage by Service Provider Current System Architecture