Bridging the Gap in Off-Road Testing
In areas such as agriculture, defense, and mining, the funda-
mentals of autonomous driving are very different from those
on public roads. Road signs, lane markings, and GPS signals
are often not available, and sensor configuration must be pas-
sive and robust. Vegetation, surface irregularities, and moving
objects in the environment trigger highly variable and unpre-
dictable scenarios. These factors push conventional testing
methods to their limits in terms of reproducibility and scope.
With the virtual validation toolchain TERRA-V, AVL has devel-
oped a product that closes this technological gap. It enables
vehicle developers to design, test, and optimize autonomous
control functions – such as perception, path planning, and
actuation – under realistic conditions and with measurable
performance feedback even in difficult offroad scenarios.
A Modular and Validated Framework
TERRA-V is a modular simulation platform built around the au-
tonomous logic structure of “Sense,” “Plan,” and “Act.” TERRA-V
integrates a high-fidelity environment simulation based on the
open-source software Carla and Unreal Engine, as well as vir-
tual sensor models and dynamic vehicle behavior. To achieve
utmost physical accuracy, vehicle dynamics models are cali-
brated using AVL VSM™ and validated against real-world test
data. These models can be further refined using data from
simulation tools, such as MATLAB/Simulink.
Operating autonomous systems in off-road environments presents unique
challenges: from difficult terrain classification and unpredictable obstacles
to reduced sensor availability. TERRA-V is AVL’s simulation-based toolchain
that enables reliable, time- and cost-efficient development of autonomous
driving functions under realistic conditions.
HOME OF INNOVATION
E-MOBILITY
Optimizing Autonomous Mobility
in Off-Road Environments
through Virtual Testing
TERRA-V
The special system architecture supports Model-in-the-Loop
(MiL), Software-in-the-Loop (SiL), and Hardware-in-the-Loop
(HiL) workflows, enabling it to be adapted to various develop-
ment stages. Via AVL Model.CONNECT™, TERRA-V allows the
seamless co-simulation of multiple components across over
30 modeling tools.
AI-Driven Testing and Measurable Results
Critical test scenarios can be generated using AI-driven, game-
based testing to simulate complex interactions with dynamic
and static objects. After each simulation run, key performance
indicators (KPIs) are used to evaluate system performance in
terms of safety, efficiency, and comfort. This structured KPI
framework allows for continuous optimization and transparent
benchmarking of all functional modules.
Designed for Rugged Applications
Its flexibility and level of detail makes TERRA-V well-suited for
a wide range of off-road use cases – from unmanned military
vehicles and agricultural automation to robotic mining equip-
ment. Simulation makes it possible to ensure performance
and safety before the first prototype hits the field.
Significant parts of TERRA-V have been developed with funds from the Austrian
defense research program FORTE. The authors would like to thank the BMF, the
BMLV, and the FFG for their support.
2025