Probabil babilisti istic c Sen ensor or Models els for Virtua ual l Val alidat dation ion – Use e Ca Cases es an and Ben enef efits its Dr. Robin obin Sch chuber bert Co Co-Founder der & C CEO BASEL SELAB ABS GmbH bH Apply & Innovate 2016 2
BASE SELABS S enabl bles es data fusi sion on results. lts. 3
Who we are What we do We partner Data fusion enthusiasts Focus on multiple sensor Vector is strategic partner with scenarios – with software 49% of shares Team of 25 engineers, products and projects software developers and Partnership with simulation managers Enable data fusion as a key providers for improved virtual technology for automated validation Software supplier to OEMs driving and automotive suppliers Active contributors to advances in research Apply & Innovate 2016 4
A A brief ef surve rvey on dat ata fusion ion Apply & Innovate 2016 5
Exampl mple of of a 36 360 ° data fusion ion sys ystem tem for automa omated ted drivi iving ng BASELABS‘ miss ssion ion is is to to provi vide de a a unified ied, , ambi biguit guity-fre free, , reliab able le repres resent ntatio ation of of the vehicle‘s environm ironment nt Apply & Innovate 2016 6
Ƹ Data fusion on in a nutshe shell A A brief summ mmary ary of of Bayesian sian filte teri ring ng and multi ti objects ts trac acking king World Model Sensor Model Sensor Data Posterior Prior Sensor data Evaluation state vector state vector hypothesis 𝑦 𝑙−1 ො 𝑦 𝑙 ො 𝑨 𝑙 Update Apply & Innovate 2016 7
Virtual ual valid idatio tion in a nutshe shell ll A brief summ A mmary ary of of model/so software ftware/hard hardware ware-in in-the he-loo oop p (xiL iL) ) vali lida datio tion DUT = World Model Sensor Model Device under Test Simulated DUT Scenario Sensor data entities Feedback Apply & Innovate 2016 8
Ƹ Data fusion on revisi isited ted Prope pertie rties s of of a a good sens nsor or mode del in Bayesian ian filtering ring - High similarity between simulated and real Sensor Model Sensor Data sensor data (in case of valid hypothesis) - Similar level of preprocessing - Mathematical sensor model needs to Sensor data Predict capture sensor behavior Evaluation hypothesis - Parameters of the model need to capture 𝑨 𝑙 properties of specific sensor device - Model needs to deal with data variability - Fast comparison between prediction and data Probabilistic sensor models (deterministic & probabilistic part) Apply & Innovate 2016 9
Exampl mple: : Sensor or mode del l for a m multi ti objects cts track cking ing sys ystem tem Mode delled led pheno nome mena na World model 3 object hypotheses with position, direction and speed D T Sensor model: We expect 3 sensor detections of which some may not be detected ( false negatives ) whereas we might get some clutter ( false positives ) and the detections may not be precisely where expected ( noise ) plus they might be delayed ( latency ) This expectation is evaluated against the actual sensor data Apply & Innovate 2016 10
Use ca cases es an and req equir ireme ements nts for virtua tual val alidation dation (Focu ocus on n simulat mulatio ion of of ADAS AS-rel relat ated ed senso nsor data) Apply & Innovate 2016 11
Use Case 1: „Lucky -Pa Path th-Testing sting “ Test of developed system under optimal conditions Rationale: „ If it does not even work with perfect data yet, I do not need to start testing with real data .“ Potential target audience: Developers of virtual proof-of-concept implementations Testers at the beginning of the test process Apply & Innovate 2016 12
Use Case 2: „Virtual Validation“ Ensure the correct functionality of the system Complement or even partial replacement of field testing Objective: Simulation should be as realistic as possible Limitation: Real-time requirements of test system (in particular for HiL-setups) Most realistic sensor models: Simulation on physical level (EM-waves) Camera images Radar power Would physical simulation models be the solution if they were fast enough? Yes and No (a.k.a. it depends) Apply & Innovate 2016 13
The required level of simulated sensor models depends on the DUT‘s input Tracks Raw Data Detections Detector/ Sensor Tracking Function Classifier Image Image Object List Example Regions of Interest Apply & Innovate 2016 14
The required level of simulated sensor models depends on the DUT‘s input Targe get t group A: Deve velo lope pers rs and test sters rs of of image ge processin sing (dete tecto ctors rs/class lassifi ifiers rs) DUT Image Detector/ Camera Classifier Input of DUT: Camera images Requirement for virtual validation: Simulated camera images that are as realistic as possible Optimal solution: Sensor models on physical level (image rendering) Apply & Innovate 2016 15
The required level of simulated sensor models depends on the DUT‘s input Targe get t group A2: Deve velop opers rs and teste ters rs of of complet lete data processin sing chain includin ding image ge process ssing ng DUT DUT DUT Tracks Image Detections Detector/ Camera Tracking Function Classifier Input of DUT: Camera images Requirement for virtual validation: Simulated camera images that are as realistic as possible Optimal solution: Sensor models on physical level (image rendering) Apply & Innovate 2016 16
The required level of simulated sensor models depends on the DUT‘s input Targe get t group B: Functio tion deve velop opers rs and test sters rs DUT ? Tracks Image Camera Function Input of DUT: Track list (e.g., list of vehicles in front of ego vehicle) Requirement for virtual validation: Track lists that are as realistic as possible Optimal solution: Sensor models on physical level (image rendering) Sensor model that behaves like a realistic tracker Smart sensors (e.g., MobilEye cameras or recent radars) do not even output raw data, just track lists. Apply & Innovate 2016 17
The required level of simulated sensor models depends on the DUT‘s input Targe get t group B2: Tracki cking/Dat ng/Data a Fusion ion deve velop opers rs and test sters rs DUT ? Raw Data Detections Sensor Tracking Input of DUT: Detections (e.g., list of image regions containing vehicles) Requirement for virtual validation: List of detections that are as realistic as possible Optimal solution: Sensor models on physical level (image rendering) Sensor model that behaves like a realistic detector/classifier Apply & Innovate 2016 18
Inte term rmedia diate te Summar mmary: y: Claims ms and Assumptio umptions ns Idealized (=error free) sensor models cannot be more than a starting point Sensor models on physical level are a valuable solution for developers and testers of systems based on sensor raw data (e.g., detectors/classifiers) Sensor models on physical level are not suitable for developers and testers of data fusion, tracking, or functions (as well as users of smart sensors) For this target group, sensor models should simulate a realistic behavior on the correct processing level Apply & Innovate 2016 19
Probabil babilisti istic sen ensor or model dels BASEL SELABS ABS Models dels for Carmake rmaker Apply & Innovate 2016 20
Ƹ There is is no no need to to re re-inv nven ent the wheel Proba babil bilistic istic sens nsor or models ls from m the tracki cking ng/da data ta fusi sion on domain ain are designe igned for these se requi uirem rement nts World Model Sensor Model Sensor Data Posterior Prior Sensor data Predict Evaluation State vector state vector hypothesis 𝑦 𝑙−1 ො 𝑦 𝑙 ො 𝑨 𝑙 Update Apply & Innovate 2016 21
BASELA LABS BS Model els s for CarMake aker (planne nned) - Probabilistic sensor models from BASELABS - Development based on extensive experience using real ADAS sensors - Models can simulate detector and tracking output - Tight integration IPG Carmaker (including visualization, parameterization, simulation, and licensing) Apply & Innovate 2016 22
Model elled led sensors ors ADAS sens nsors ors with typical al interfaces rfaces Radar Detector Provides an interface typical for automotive radar sensors • Delivers measurements in Polar coordinates relative to the sensor position • Models specific errors of radar sensors delivering detections • Smart Camera Provides an interface typical for automotive smart camera systems • Delivers measurements in Cartesian coordinates relative to the sensor position • Errors common to sensors Latency • Apply & Innovate 2016 23
Recommend
More recommend
