aiDrive™

Automated driving platform for all levels of autonomy

 

Our perception solution – MS²N

State-of-the-art perception serves as the foundation of all automated driving solutions. All aiDrive implementations are built on our Multi-Sensor Model-Space Neural Network (MS²N), ensuring that we deliver the safest and most robust solutions on the market.

Data-driven

Delivers data-driven perception and fusion based on multiple sensor modalities 

Model-space

Provides more clarity and higher robustness in the first step of perception by working in model-space 

Robustness

Has immediate time consistency information, increasing tracking robustness

Superior performance

Achieves superior detection performance, especially in challenging ODDs

Reusable

Is less sensor dependent, allows for maximum reuse of collected and annotated data 

Reference solutions

L2+ Highway Assist Application 

The aiMotive fleet is racking up miles of real-world test driving every day in both Europe and the US, thanks to constant improvement to the features, performance and robustness of our Highway Assist solution. Using the latest available camera and radar sensors as well as navigational map support, aiMotive can create state-of-the-art Highway Assist solutions with Automatic Lane Change and Highway Interchange Management or provide specific software modules to complement our customers' systems.

L2+ highway assist application sensor visualization
 

Specifications

Minimum sensor set 6x min. Full HD camera 70-190° FoV, 1x long-range radar, 4 x corner radar
Supported features ACC, LKA, FCW, AEB, LDW, LCDAS, ISA, TSR, TLR, Blind Spot Assist, Automated Lane Change, Highway Interchange Management
Perception features Lane and road marking detection
Object detection and classification: vehicles & VRU-s, traffic signs, traffic lights
Fusion feature Road model
MS²N early fusion
Localization features Vehicle dynamics-based odometry
Lane-level localization
Motion planning Route planning
Behavior planning
Trajectory planning
Target platform NVIDIA Xavier with QNX, Qualcomm Snapdragon Ride
 

L3 Highway Chauffeur Application

This solution expands over the currently available regulatory boundaries to achieve the same functionality as the L2+ Highway Pilot feature (including ALCS and HIM) with a higher level of automation. The system is capable of executing a minimum risk maneuver if the driver does not take control in the required timeframe.

L3 highway chauffeur application sensor visualization
 

Specifications

Minimum sensor set 11x min. Full HD camera 70-190° FoV, 2 x long-range radar, 4 x corner radar, 4 Lidar
Supported features ACC, LKA, FCW, AEB, LDW, LCDAS, ISA, TSR, TLR Blind Spot Assist, Unsupervised Lane Change, Highway Interchange Management
Perception features Drivable free space detection
Lane and road marking detection
Object detection and classification: vehicles & VRU-s, traffic signs, traffic lights
Fusion feature Road model
MS²N early fusion
Localization features Vehicle dynamics-based odometry
HD map-based localization
Motion planning Route planning
Behavior planning
Trajectory planning
Target platform Dual NVIDIA Xavier with QNX, Dual Qualcomm Snapdragon Ride
 

L3 Home Parking Application

This feature spares the car owner from having to drive in and out of their driveway or garage at home. The vehicle can learn a fixed route to navigate into its parking space and carry out that task at any time of the day or night, only requiring external supervision from the driver. 

L3 home parking application sensor visualization
 

Specifications

Sensor input 4 x min. Full HD camera > 180° FoV, 4 x corner radar, 12 x USS system
Supported features Navigating a pre-recorded route into driveway or home garage
Perception features Drivable free space detection
Object detection and classification: vehicles & VRU-s
Fusion features Occupancy grid fusion
Localization features Visual odometry
Vehicle signal-based odometry
Visual landmark-based localization
Motion Planning feature Global route-based navigation
Target platform TI TDA4, NVIDIA Xavier with QNX, Qualcomm Snapdragon Ride
 

L4 Automated Valet Parking Application

Parking is undoubtedly a time-consuming and tedious task for most drivers, especially in complex parking garages. aiMotive's Automated Valet Parking paves the way for the next generation of parking solutions, enabling vehicles to move through and park in structured parking environments without the driver being in or near the car. Our solution relies on in-vehicle intelligence instead of expensive investments on parking infrastructure and runs on embedded automotive SoC-s.

L4 automated valet parking application sensor visualization
 

Specifications

Sensor input 4 x min. Full HD camera > 180° FoV, 4 x corner radar, 12 x USS system, 1x Lidar 
Supported features Driverless vehicle maneuvering in structured parking facilities, driverless parking-in and parking-out maneuvers, forward and backward AEB
Perception features Drivable free space detection
Object detection and classification: vehicles & VRU-s, road markings, traffic signs
Parking space detection
Fusion features Occupancy grid fusion
MS²N early fusion
Localization features Visual odometry
Vehicle signal-based odometry
Visual landmark-based localization
Motion Planning features Parking structure path planning
Parking space maneuvering
Target platform TI TDA4, NVIDIA Xavier with QNX, Qualcomm Snapdragon Ride
 

Front camera-based ADAS

Our first and foremost priority is to provide technologies that increase safety on roads. In addition to higher automation level comfort features, we also offer a wide range of ADAS features to our customers, allowing them to comply with regulations, as well as add value to their vehicles in the eyes of their buyers.

front camera-based ADAS sensor visualization
 

Specifications

Sensor input Front min. Full HD camera, Front LRR
Supported features ACC, AEB, FCW, LDW, LKA, TSR, TLR
Perception features Lane and road marking detection
Object detection and classification: vehicles & VRU-s, traffic signs, traffic lights
Fusion features MS²N early fusion
Road model
Localization features Visual odometry
Vehicle signal-based odometry
Motion Planning features –
Target platform TI TDA4, NVIDIA Xavier with QNX, Nextchip Apache5 and Apache6
Interested in meeting challenges together?

Get in touch with us

Our team is always ready to work with exciting and ambitious clients. If you're ready to start your partnership with us, get in touch.

Contact Us