There used to be a huge gap between simulation and reality…
Autonomous systems would be painstakingly tweaked and tuned in software…only to fail on day one of real world testing. It didn’t matter if you were flying a drone, driving a ground rover or piloting an autonomous surface vessel.
Operators faced the same problem.
Simulation said X would happen. Reality said Y happened.
Fortunately:
Hardware components are quickly closing that gap. Advanced autopilot controllers, sensor packages, hardware-in-the-loop testing rigs; unmanned vehicle components are empowering teams to test in the virtual world and deploy in the physical with confidence.
Here’s the breakdown.
What you’ll discover:
- Why the Simulation-to-Reality Gap Is Important
- Autopilot Controllers That Work In The Sim & Real World
- Hardware-In-The-Loop Testing Hardware
- Sensors To Bridge The Gap Between Virtual & Physical
- Communication & Telemetry Hardware
- How To Choose The Right Parts
Why the Simulation-to-Reality Gap Is Important
Think about sports for a minute.
Practice is the simulation. Game day is reality.
If you don’t use the same gloves for practice as you do in the game, every throw becomes suspect. You just don’t know how the equipment will perform until it’s show time.
Same concept applies here.
If unmanned vehicle operators aren’t using hardware components in simulation that can be deployed in real world applications, then no matter how well the test cases perform, there’s no guarantee those scenarios will translate to success on game day.
And the market isn’t slowing down anytime soon either. The UAV market size alone is projected to reach $42.39 billion by 2025. Throw in unmanned ground vehicles, autonomous surface vessels and underwater robots and you’re looking at billions of dollars of hardware and software ripe for the taking.
That’s why many of the companies acting as a trusted Cube Pilot partner are so vital. They provide autopilot hardware and unmanned vehicle components that can be used in both worlds. When your controller is capable of running the exact same mission in a virtual test bench as it can in the real world, the simulation-to-reality gap starts to close.
The bottom line?– All aspects of unmanned vehicle testing should take hardware into consideration. Air, land or sea.
Autopilot Controllers That Work In The Sim & Real World
You can’t have an unmanned vehicle without an autopilot. Simple as that.
It doesn’t matter if it’s a multi-rotor flying above farmland, a ground rover on wheels zipping through a warehouse or a surface vessel scraping along the ocean floor. Someone has to make it go!
That “someone” is usually the autopilot.
The best autopilot controllers have several things in common:
- Compatibility with leading simulation software ecosystems like Gazebo and AirSim
- Support for open-source flight firmware such as ArduPilot or PX4
- Ability to run the same software stack in simulation and deployment scenarios
CubePilot controllers fit that criteria and then some. Because the hardware is built to integrate with popular simulation software packages, you can run the exact same flight mission inside the simulator your vehicle will run outside.
No code changes. No guess work. Just rock solid reliability.
Hardware-In-The-Loop Testing Hardware
Ok so hardware-in-the-loop (HIL) testing may sound intimidating, but it’s honestly one of the simplest concepts in unmanned vehicles.
Essentially, HIL testing bridges the gap between pure software simulation and physical testing by connecting real autopilot hardware to a virtual environment. The controller is fed data as if it were on a real mission, but every input and output comes from software.
This can be applied to any unmanned platform:
- Ground vehicle autopilots can be connected to navigate through simulated waypoints
- Surface vessel controllers can be tested against virtual ocean swells and waves
- Aerial autopilots can evaluate how simulated wind affects their control algorithms
HIL is great at catching bugs that software-only testing may miss. Introducing real world hardware adds complexity like code compile times, sensor noise and electrical interference. Capturing these variables in testing means more successful deployments in the field.
Sensors To Bridge The Gap Between Virtual & Physical
Alright, say you have the perfect autopilot controller. It runs the same flight stack in Gazebo as it does on your actual drone.
But now what?
Your drone still needs sensors to see and hear what’s going on around it.
Thankfully modern unmanned vehicles come packed with a suite of sensors that translate well between virtual and physical applications:
- GPS / RTK modules can emulate outdoor positioning data
- LiDAR sensors provide virtual maps and points of interest
- Inertial measurement units can be tested inside simulation software
- Sonar / Depth finders help bridge the gap for surface vessels and underwater robotics
All of these sensors provide live input to the autopilot. Without accurate sensor data, your unmanned vehicle is essentially flying blind.
But here’s the kicker. Not only are these sensors critical for real world testing, they can all be emulated inside simulation software. GPS coordinates, LiDAR point clouds, sonar pings, camera video feed…you name it. Simulation software like Gazebo can emulate it.
When the virtual sensors match reality, tuning your autopilot in simulation becomes that much more effective.
Communication & Telemetry Hardware
Here’s a wildcard that flies under the radar a lot of the time…
Communication and telemetry.
As important as the autopilot and sensors are, teams need to be able to see what’s going on in real time if they want complete awareness of their unmanned vehicles.
That’s where long-range radio telemetry gadgets come in. Whether it’s for beyond visual line of sight communications, streaming realtime data or securing your control link with encryption, these systems are just as important to test in simulation as anything else.
Telemetry radios that work flawlessly during a bench test could easily falter in the field. Keep that in mind.
Oh and by the way, there are now over 855,000 drones registered with the FAA. Unmanned ground vehicles are taking off and marine robotics are picking up steam.
Anyone who expected “remotely-piloted” vehicles to replace manned counterparts likely didn’t see that kind of growth coming.
Telemetry and communication gadgets are no longer an afterthought.
How To Choose The Right Parts
Picking the right unmanned vehicle components can be tricky.
But if you keep these tips in mind, you’ll be setting yourself up for success.
First off, make sure the hardware you choose can be used in simulation. Does the autopilot support HIL testing? Can your sensors be emulated? If the answer is no, you’ve got a long way to go before closing the simulation-to-reality gap.
Next, look for components that will work on more than one platform. Being able to fly, drive and sail with the same ecosystem of hardware saves time and money.
After that, durability is key. Simulation testing doesn’t face the same weather, vibration or extreme temps that real-world deployments do. Make sure your gadgets can handle it.
Lastly, try to pick parts that are easily scalable. Can the hardware grow with your fleet? Will additions need to be made from scratch or does the ecosystem support modular designs?
Wrap Up
Finding the right unmanned vehicle components to bridge the gap between simulation and reality is no easy task. But now that more gadgets are built with testing in both environments in mind, teams have more options than ever before.
Just remember these key points and you’ll be set:
- Make sure your autopilot controller can run your mission in both worlds
- HIL testing is your friend
- Sensor and telemetry gadgets should be mirrored in simulation
- Picking durable, scalable components that work with your simulation software will save you headaches down the road
Unmanned vehicles are taking over skies, land and sea. Whoever invests in the right components now will lead the pack in years to come.
