SAFEAI Technical Paper Achieving Safe and Swift Autonomy With SAF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

TECHNICAL OVERVIEW

ACHIEVING SAFE AND


SWIFT AUTONOMY
WITH SAF
Brought to you by SafeAI
EXECUTIVE OVERVIEW
The possibilities for autonomous technology are rapidly increasing,
and organizations across all industries are eager to adopt autonomy
into their business model.

The astute business leader understands that Business leaders must face choices on
a delicate balance between speed and safety updating their equipment and systems, rising
must be achieved and maintained before full- to meet increasing expectations in safety and
scale implementation of autonomy at any sustainability, and identifying flexible and
level can be considered successful. This scalable solutions—all while trying to match
balance is especially critical for heavy the pace of autonomy’s evolution. While these
industries, where the potential and severity barriers to entry into autonomy may be
for harm is greatly amplified. significant, we believe that they are not
impossible to scale. In fact, we believe that
Industries that necessitate a risk-averse safety and speed can be balanced without
approach, such as mining or construction, compromise so that companies can achieve
face certain challenges. autonomous integration even faster and with
greater success.

The SafeAI Autonomous Framework (SAF) is our solution to these


challenges.

Our framework has been purpose-built to meet the unique needs for autonomy
in heavy industry:

• Extremely reliable
• Self-sustaining
• Robust
• Highly versatile

With SAF’s safety-certified status, not We are confident that SAF can save
only can you be sure that safety has been you invaluable time and money as you
incorporated into every aspect of our venture into the future of autonomous
product’s development, but you can also implementation and operation.
expect that same level of safety rigor for
any of your applications built upon SAF.
NAVIGATION

TABLE OF CONTENTS

Executive Overview ............................................................................2

Introduction...........................................................................................4

Addressing Challenges in Autonomy for Heavy Industry...........5


Investing in equipment and systems
Meeting increased expectations
Finding flexible and scalable solutions
Catching up with autonomy’s evolution
SafeAI’s solution

All About SAF........................................................................................ 7


Main Features
Architecture Overview
Accelerating Functional Safety
Safety in the development process
Safety and quality at SafeAI
Safety certification
Current Support

The Wrap Up.........................................................................................9

3 AC H I E V I N G S A F E A N D S W I F T AU TO N O M Y W I T H S A F
LET’S GET STARTED

INTRODUCTION
Heavy industry companies recognize autonomous technology’s potential, but face
significant barriers to entry in the form of limited time, few resources, and
a lack of in-house expertise to build these solutions. That’s why we created the
SafeAI Autonomous Framework (SAF), the first operating system purpose-built
for autonomous heavy equipment.

As the possibilities for autonomous technology are


rapidly increasing, organizations across all industries
are identifying new opportunities to adopt autonomy
into their business model. At the same time, each new
advancement in autonomy creates ebb and flow in the
tide of public opinion with regard to safety and
responsibility.

Strong business leaders will undoubtedly keep a


pulse on these shifts, and understand that a delicate
balance between speed and safety must be achieved
and maintained before full-scale implementation of
autonomy at any level can be considered successful.

Maintaining this balance is especially critical for


industries like mining and construction. The nature of
worksite operations and the size of the machinery
involved greatly amplify potential and severity of
harm stemming from unaddressed hazards. It goes
without saying that work sites must prioritize a risk-
averse approach when human lives are on the line.
Comparing this with the full-throttle pace at which
businesses are expected to run in order to remain
competitive, it is easy to conclude that these two
aspects conflict to a degree where significant
concessions or sacrifices must be made.

SafeAI believes that implementing autonomous


technology in a way that perfectly balances safety
and speed without compromise to either is not only
possible, but achievable today. This technical
overview describes how the SafeAI Autonomous
Platform (“SAF”) can provide your organization with a By providing developers access to
solid foundation for your software developers to build SafeAI’s certified and production-
safety-compliant applications and environments and
bring your company up to speed regardless of your ready infrastructure, SAF helps
current stage on the autonomy journey. companies fast track their
autonomous development and
deployment, and experience the
benefits of autonomy faster than
ever and at a lower barrier to entry.
4 AC H I E V I N G S A F E A N D S W I F T AU TO N O M Y W I T H S A F
M AKI NG
AUTONOM Y
M ORE
ACCE S SI BLE

ADDRESSING CHALLENGES IN
AUTONOMY FOR HEAVY INDUSTRY
Since SAF is use case agnostic, companies across heavy industry can build off
of this framework, and create customized applications to suit their specific needs
and use cases. SAF is to heavy industry what Android is to mobile.

Headlines about the futuristic “driverless passenger Of course, the path to off-road autonomy is not so clear-
car” may effortlessly catch the public eye, but many cut for many companies: there are many barriers to entry
industries are recognizing that autonomous vehicular that prevent an easy transition to full autonomous
technology holds potential that extends well past the operation. What follows are some of the most common
on-road use case. challenges that businesses face when it comes to
successful implementation of autonomous software in
If properly integrated, a worksite using autonomous their operations.
equipment can expect greater operational efficiency
and lowered risks for mining personnel. Automation
also allows for more precise monitoring and control of Investing in equipment and systems
equipment use and fuel consumption, leading to
subsequently lowered costs on maintenance and fuel. While fully autonomous heavy industry vehicles currently
All of these variables can result in significant projects exist on the market, even one of these new vehicles can
being completed in a shorter amount of time and at a cost upwards of $4-6 million USD alone. While larger
lower budget than expected. corporations with market capitalizations in the billions will
have no trouble paying for full-scale upgrades up-front,
smaller enterprises may struggle to rationalize investing in
one of these new vehicles, let alone replacing their entire
Even one autonomous vehicle can add fleet. This problem is compounded even further when
approximately 1,000 additional hours adding locations of operation across different countries.
Careless implementation of anything can create
of productivity per year. unexpected costs for an organization through global
supply chain disruptions, and autonomy is no exception.

5 AC H I E V I N G S A F E A N D S W I F T AU TO N O M Y W I T H S A F
Our framework solution has been
Meeting increased expectations
purposefully built to address challenges
Those who work with heavy machinery know for the heavy equipment industry, as
that implementing autonomy is not as simple as
wheeling in a new truck to a work site. The use
well as relevant safety expectations.
of just one autonomous vehicle will require a With SAF, developers can immediately
complete reassessment of existing infrastructure
and systems to reassure safety at every stage.
begin working in a production-ready
These expectations will translate to updated environment to create safety-certifiable
responsibilities and processes for all employees,
and new skills may be needed to fill in new gaps
custom functions and applications.
in expertise. For software developers, these duties
may require increased education, purchasing
access to new standards, and even training to
the level of proper certification.

Talks of safety for autonomy in certain industries Furthermore, autonomy is not something that is
also often go hand-in-hand with discussions of statically implemented at one singular point.
sustainability and environmental stewardship. Since Businesses change and evolve and ideally, your
efficient fuel usage and lower emissions are often autonomous solution should be able to adapt to
selling points for autonomous vehicles, many parties match the pace and growth of your business needs.
are looking to sites that implement autonomous
technology to gauge whether it can be a significant
factor in preserving our future.
Catching up with autonomy’s
evolution
Finding flexible and scalable All of the aforementioned challenges are further
solutions complicated with the speed and timing of autonomy’s
evolution. Teams may find themselves pressured
While autonomy has its benefits for any company’s to quickly get their operations up to speed in order
operations, no two businesses will share the same use to reap maximum benefits of being on autonomy’s
cases. In any industry testing the waters of autonomy, cutting edge. Access to the latest developments
businesses may be at vastly different stages of can mean identifying and acting upon new business
implementation due to factors like size, available opportunities and innovations that much faster.
internal resources, customer base, and existing However, new software can realistically only be
progress. A company that is already up to speed in developed at the pace of your team’s shared
terms of their software toolchain will require a different resources and knowledge. It may seem impossible
solution from a company that is just getting started for many to sufficiently meet these new expectations
and essentially playing “catch-up”. in safety and security while maintaining a
competitive pace in the market.

SafeAI’s solution
Issues with deploying autonomy are significant, but not insurmountable. We believe that SAF can provide
businesses in heavy industry a chance to not only break through these barriers to entry, but to break ahead.
By providing developers access to SafeAI’s certified and production-ready infrastructure, SAF can alleviate the
barriers most companies face and save companies valuable time — around 4 years — by greatly accelerating your
development progress for smart applications for heavy industry. SAF also provides developers with a ready-built
software framework to build from, enabling more companies than ever to roll out their own autonomous
applications, while also saving millions of dollars in development costs.

6 AC H I E V I N G S A F E A N D S W I F T AU TO N O M Y W I T H S A F
THE KEYS TO AUTONOMY

ALL ABOUT SAF


SAF is the world’s first comprehensive and safety-certifiable platform that has been
designed to suit the needs of heavy equipment and machinery. The main goals for
SAF can be summarized as follows:

To be highly reliable To be self-sustaining To be robust To be versatile

Main Features
Our solution boasts many features that support these • To be robust, SAF supports a wide variety of
four goals. As an OS-isolating framework, SAF wraps complex features necessary for autonomy
the complexities of any underlying operating system operation and management such as inter-task
and relevant POSIX calls into a high-quality and and inter-process communications, multitasking,
developer-friendly API. This allows retrofitted task synchronization, and automatic console-
hardware to interact seamlessly with SafeAI’s and file-logging mechanisms.
autonomy drivers and libraries. Additional functions
like automatic console- and file-logging mechanisms • To be versatile, SAF has been made to have a small
and integration with the native capabilities of individual footprint and be modular by design. You
commonly used third-party libraries such as ROS2 are able to use SAF independently or connect its
make SAF an easily adaptable tool for your team’s processes with features from third-party libraries
specific needs. such as ROS2.

Let’s take a closer look at how SAF accomplishes For software developers working to implement
these goals. autonomous solutions in the heavy industry, SAF
can be the answer to many of their challenges.
• To be highly reliable, our framework has been
designed and built from the ground up with the
highest functional safety expectations and industry Architecture Overview
guidelines in mind. By using this framework as a
foundation, developers can expect that any SAF is a framework that consists of five parts:
applications built upon SAF will also have the same
• A System Isolator that allows app code to remain
level of compliance and rigor in their final product.
the same regardless of the original operating
system or compilers that were used for the build;
• To be self-sustaining, SAF offers OS and
communication library isolation, so your app code
• A C++ Wrapper that allows developers to quickly
is not changed regardless of which OS you are
write code that is compliant with common coding
currently using. Therefore, you are not “locked in”
guidelines and safety standards for use in
to using any specific OS, compiler, or vendor.
embedded and real-time environments;

7 AC H I E V I N G S A F E A N D S W I F T AU TO N O M Y W I T H S A F
• A Non-Deterministic System Isolator that Safety in the development process
provides an interface to inherently non-
deterministic TCP/IP API; Many standards provide guidance on ensuring
functional safety within the software development
• A Non-Deterministic C++ Wrapper that provides and testing process. SAF has been developed to
a high-level interface to the non-deterministic comply with these expectations, as well as specific
system isolator; and coding guidelines established by MISRA and
AUTOSAR for software in the automotive industry.
• A Light Non-Deterministic C++ Wrapper for ROS2
that provides waiting calls to ROS2 objects. We run a thorough test suite for each new
development before publishing anything to release.
SAF bridges the gap between the application layer These tests utilize in-depth modified condition/
and the real-time operating system to provide a decision coverage (MC/DC) for 100% code line
unified interface experience for users. coverage, and will detect and flag rule violations to
MISRA and AUTOSAR
coding standards. The
presence of any violations
or unexpected outputs
require development to
halt and remedy the error
before continuing into
production.

SAF is also run and


intensely tested on
hardware and operating
systems that have been
pre-certified for use in
safety applications. More
information about these
testing environments can
be found in SAF’s safety
package.
Figure 1: Architecture model of the SafeAI Autonomous Framework

Safety and quality at SafeAI

Accelerating Functional Safety A product’s status as a “functionally safe”


component depends on much more than its internal
SafeAI focuses on delivering safety as a priority in all construction. Safety must be present throughout the
of its products, and SAF is no exception to this rule. organization’s environment and culture as a core
Like all SafeAI products, SAF is continuously tenet in order for a product’s safety claim to be
developed to comply with the latest safety reliable. The organization must also have processes
expectations in the industry. To further secure our in place to ensure that a product has considered and
claim as a safety-certifiable solution, SafeAI has implemented safety from its inception through its
undergone the rigorous functional safety certification distribution and continued use. The individuals
process to verify that SAF is capable for use in safety- working on the product must also prove themselves
related applications. to be competent and sufficiently trained. SafeAI

8 AC H I E V I N G S A F E A N D S W I F T AU TO N O M Y W I T H S A F
models its processes on common frameworks designed Current Support
for quality management such as ISO 9001, IATF 16949,
and ASPICE. Our company also follows and adapts to the
SAF currently supports the following operating
guidance of prominent industry voices on the local,
systems and architectures:
national, and international levels.

x86_64 ARM 64
Safety certification
QNX Yes Yes
Linux Yes Yes
We have worked closely with TÜV NORD, an
internationally recognized partner for test and Linux RT Preempt Yes Yes

certification services, to accomplish this feat and ensure Windows Yes On demand only

quality in our product’s audit. Apple OSX On demand only Yes


iOS - On demand only
Our current certification is for compliance with ISO BSD On demand only On demand only
26262, which is itself a well-defined automotive VxWorks On demand only On demand only
specification of the general IEC 61508 standard for
electrical and electronic safety-related systems.
SAF also provides support for the GCC/G++,
Throughout our certification process, we have tailored
Microsoft Visual C/C++, and Apple CLang C++
our safety process and safety case to incorporate our
(XCode) compilers. Additional support can be
specific use cases and assumptions. The parallels we
provided upon request.
have established between ISO 26262 and IEC 61508, as
well as other applicable standards for our industry, have
been noted in our safety documentation. We can state
with confidence that alignment with these other
standards is entirely achievable.

CONCLUSION

THE WRAP UP
Developers utilizing SAF can bring the benefits of autonomous operations to
their products faster than ever, at a lower barrier of entry, and with the highest
safety standards.

Your business can save time and money with a safety-certified solution that guarantees compliance with the most
applicable standards. We believe companies looking to develop their own autonomous solutions can achieve great
things using SAF in their workflow.

We understand that everyone’s autonomy needs are different, and we are excited to see how SAF can help you.
For customers who choose to adopt SAF, the SafeAI team will also provide general engineering and troubleshooting
support as needed. Please contact us if you are interested in learning more, or in obtaining a trial license for SAF.

9 AC H I E V I N G S A F E A N D S W I F T AU TO N O M Y W I T H S A F
ABOUT

SafeAI is powered by a seasoned team with deep knowledge


of construction and mining industry operations, two decades of
off-road autonomous vehicle production experience, and expertise
in the development and implementation of AI powered technology.

Through a collaborative ecosystem of customers, partners and
resources, we are successfully scaling our autonomous solutions
for leading mining and construction companies across the globe.

www.safeai.ai

SafeAI USA
5201 Great America Pkwy,
Suite 332 Santa Clara, CA 95054

SafeAI Australia
Central Park, Level 32, 152 St Georges Tce
Perth WA 6000

SafeAI Japan
Ginza Daiei Building 5F,
1-16-7, Ginza, Chuo-ku, Tokyo 104-0061
E-FS-ALL-5-0002 – VERSION 2.00 04.2023

SafeAI India
WeWork India Management Private Limited
Block L, Embassy Tech Village
Bengaluru, KA 560103, India
Office No - 158, 163

© Copyright 2023 SafeAI. All rights reserved.


Specifications subject to change without notice.

You might also like