Page 17 - EE Times Europe Magazine – June 2024

EE|Times EUROPE 17

AUTONOMOUS VEHICLES | SENSORS
How Versatile Technologies Can Boost

AV Adoption

By Saumitra Jagdale

he autonomous vehicle (AV) industry is forging ahead despite divided tion of their surroundings. At some point,
opinions on their adoption, technological advances, reliability and AVs will be subjected to harsh environmental
conditions like snow and rain, which can alter
feasibility. Statista estimates that about 54 million cars across the the data collected by the sensors, leading to
T globe will have some degree of autonomy by the end of 2024. How- accidents. As part of a test, Innoviz subjected
its sensors to over 120 hours of driving in
ever, this has only heightened concerns over the safety of passengers in AVs,
10 countries. But as indispensable as testing
and the consensus over the unreliability of AVs directly hampers their adoption. may be for the sensor itself, “InnovizTwo is
The industry’s response has been to improve automated driving systems— designed as a platform rather than a single
configurable product,” Elad Hofstetter, CBO
including hardware components like sensors, control systems, on-board of Innoviz Technologies, said in an interview
computers and communication systems, as well as software algorithms encom- with EE Times Europe.
passing perception, localization, mapping, planning, decision-making, machine The data collected by InnovizTwo sensors
is used to improve Innoviz’s proprietary
learning and artificial intelligence—to enable vehicles to operate autonomously AI-powered perception software. “The
or semi-autonomously. hardware development team works hand in
hand with the software team to ensure that
While advancements in software and hardware components are the LiDAR design aligns with the requirements of the perception
crucial, ensuring the reliability of hardware components takes prece- software,” Hofstetter said. “For instance, the hardware team may
dence. It is therefore imperative that they undergo rigorous testing modify LiDAR components to improve data quality or increase scan-
to guarantee the accuracy of the data collected by these components. ning capabilities based on feedback from the software team regarding
Innoviz, a Tier 1 supplier of automotive-grade LiDAR sensors, launched perception algorithm requirements. Conversely, the software team
its InnovizTwo LiDAR sensors and demonstrated the sensors’ resilience provides valuable insights to the hardware team by identifying poten-
in a successful winter test. tial issues or limitations in LiDAR’s performance that may impact
perception algorithms.”
LiDAR SENSORS PAIRED WITH AI-POWERED PERCEPTION This collaborative approach enables iterative improvements to hard-
SOFTWARE ware and software components, resulting in a more robust and reliable
LiDAR sensors, which use laser beams to measure precise distances and solution. Furthermore, the perception software has calibration and
movement in an environment in real time, give AVs a precise percep- semantic point-cloud (Figure 1) features to enhance the functionality

Figure 1: InnovizTwo point cloud

www.eetimes.eu | JUNE 2024