Oztech MOVAL
V&V and Testing Framework

oztech_moval.png

Architecture of the Oztech MOVAL Testing Framework

Why oztech moval ?

The Oztech model-based validation environment (MOVAL) provides you with an approach for the software/hardware verification and testing of autonomous vehicle designs. This includes a suite of customized layouts and scenarios, vehicles, and sensors and an initial simulation testbed in Oztech’s custom MODE simulator.

The approach is risk analysis based and allows the user to track down the initial cause leading to an eventual incident.

features

Customizability

Specific environments can be modeled, and different scenarios can be imported into the MOVAL framework. Oztech’s simulation environment includes a Gazebo core, ROS-based that allows for co-simulation with MATLAB, Simulink, and SUMO for general traffic simulation.

Participant Modeling

Different cars, trucks and busses can be included, at the desired level of detail in the basic simulator for multibody. We can also add one or more pedestrians with different occlusion or emergence probabilities.

Standards Compliance

The design of MOVAL complies with intended functionality and hazard standards: HARA, SOTIF, PHA, FHA. 

Scenario Specification

Different Scenarios can be specified with our special interface and tested for risk.

Portfolio

Basic Configurations and Scenarios

  • Pedestrian crossing straight road

  • 4-way intersection

  • Truck following on highway

  • Lane change/merge

  • T-junction

Peer reviewed

Oztech’s MOVAL framework and use cases has been submitted and presented in several research venues.

Dolores & Decentralized Control Architecture

Dolores2-pic.png

DOLORES-2

Dolores 2.0 is a software package for the synthesis and design of feedback controllers for centralized and decentralized systems. It is developed in MATLAB. The software works with a wide range of cyber-physical systems area, including smart grids. It has an easy-to-use interface and deals with decentralized control system feedback problems.

Dolores-2-pic-2.png

Our Pedestrian Work

As Oztech, we focus on safety and comfort of all traffic users, but specifically Vulnerable Road Users (VRU) during model validation in our tests. 

We are actively researching on pedestrian safety, specifically for the challenging scenarios even for the humans.

These scenarios include unprotected turns, pedestrians emerging from occlusions.

Simulation Results for The Smart Pole Project

The aim of the project was analyzing the risk in Vehicle-Pedestrian Interaction with Smart Pole (external pedestrian alarm system) in Vehicle Right Turning Scenario to verify the risk of colliding with a pedestrian in an occluded right turn scenario can be reduced with an external pedestrian alarm system and to identify the most critical attribute that leads to potential collisions.

zhitong_pedestrian_work_edited.jpg
Without the Smart Pole
With the Smart Pole

Using Visible Cues for Risk Assessment under Occlusions

The aim of the project was to develop an algorithm to better tackle the scenarios with possible occluded pedestrians by using the visible cues around (parked cars, pedestrian crossings, and other visible pedestrians/VRUs).

my_thesis_summary.png

Our proposal is that the existance of the visible pedestrians and parked cars should indicate an elevated risk level between parked cars so that the AV could understand to slow down when there are occluded regions.

Novel Approach of Risk Zones

emergence_probability.png

The second novelty of the proposed method is that by dividing the future trajectory into danger, discomfort, and safety zones the AV can behave differently to the same level of risk in different regions. (e.g., a level of risk 3 seconds ahead may not be dangerous; however, it could be very dangerous if it is  just 1 seconds ahead 

Results

Proposed Method
Proposed Method
Baseline  3
Baseline  1
result_quantitative.png

The simulation results show that our proposed method outperforms the three baseline methods.

 

Our Previous Work

We have been working on automation of various devices and vehicles; with various companies including Ford, Leidos, Yazaki, and others.

COTA-ridefinderpic.jpeg
COTA-busstoppic.jpeg

01

The COTA Ride Finder

A set of busses would communicate their location to the system which, in turn, would estimate their arrival time at the hotels and at the airport. This information would be displayed at kiosks and some bus stops, together with other city-related information, both visually and vocally.

The COTA Ride Finder System was developed, installed, and tested for COTA in Columbus, Ohio. It was a Bus Location System for a special route established to connect a set of Columbus downtown hotels to Columbus International Airport.

02

Oztech Inc. can establish connections and activity between overseas companies and research centers and their US counterparts in its domain of expertise. Above is an example Project between ASELSAN Inc., Ankara, Turkey, and a US University, to develop multiple, communicating autonomous outdoor robots. The robots collaborate to jointly map an unknown area. The system was developed to compete in the international robotics competition in Australia, titled MAGIC.

autonomousvehiclespic1-681x1024.jpeg
autonomous-outdoor-robots-pic2-1024x768.jpeg
Car Lights

03

Oztech Inc. has helped a number of companies in developing an overview of technology, and completing publishable reports on their research. Recent examples include reports on:

  •  Vehicular displays

  • Traffic Speed Harmonization

  • Lane change and merge experimentation for collaborating vehicles

  •  An automated city-wide transportation fare collection system for Izmir, Turkey.

These projects were sponsored by various organizations including Ford, Leidos, Yazaki, and others.