Quanser's Journey to Engineer Unmanned Vehicle Systems for Academic Advancement

Recently, Quanser introduced a completely new platform for teaching and research - the Unmanned Vehicle Systems (UVS Lab). UVS Lab is the result of more than five years of Quanser’s internal research and development efforts. To learn more about their journey, we talked with the two people driving the innovation - Dr. Jacob Apkarian, Quanser’s Founder and CTO and Cameron Fulford, Systems and Control Engineering Manager.

Jacob: Ever since I was a kid, I wanted to fly my own devices - I guess that's where it all has roots. The unmanned research at Quanser started about five years ago when we began thinking how we can use our Quarc control system design software (or WinCon at the time) to actually fly things. Quarc's real-time capabilities allowed us to collect data online, tune parameters online, so if you have any flight dynamics issues that you want to do online, you can do it right away with this rapid prototyping software. Alas, the technology to do this was not there - so we developed it! Now I feel like an ace when flying these things - something I could never do without Quarc.

Cameron: For years Quanser offered the 2 DOF Helicopter, 3 DOF Helicopter and a 3 DOF Hover, which are not flying vehicles, but experiments simulating flying vehicles. So moving towards vehicles actually flying seemed like a natural progression.

Jacob: As engineers at Quanser started to look into the area, we realized there are a lot of people doing research in unmanned autonomous vehicles. We approached the Defense Research and Development Canada (DRDC), where we found Dr. Camille-Alain Rabbath, a scientist who supported this kind of research. We started to collaborate with DRDC on development of the vehicle control systems, but we still did not have any vehicles. We came up with the ALTAV - an almost lighter than air blimp, because we thought it would be easier to fly than something that is heavier than air.

Cameron: The first version of ALTAV had a shape of a round beach ball, which we eventually replaced with 11 and 13' long, more traditional blimp-looking shape. These ALTAV versions were all helium-filled blimp balloons with 4 actuated motors, so you could actually tilt the motors and get a vectored thrust, controlling the direction of the thrust of each motor. That was a novelty behind the design of our ALTAV vehicle.

One of Quanser's ALTAV models with its design and engineering team

Jacob: On our journey we learnt that a vehicle with four motors is a good design. As we continued our research, we discovered a lot about the IMU needed, and the sensors needed to fly these things properly. So as the time went by, we started developing the ultimate IMU board, which we now call the HiQ. From there our software engineers enhanced Quarc to generate code for the HiQ and control any vehicle. However as we started flying the ALTAV outside, we quickly realized the challenges: one of them being legal issues, because you have to get all the licenses and permits to fly outside. The other challenge was the weather.

Cameron: The ALTAV was so large that it was susceptible to winds. If there was any matter of wind, it could really affect the performance of that vehicle, so it was difficult to fly outdoors, but at the same time, was too big to fly indoors. A lot of people we are talking to ultimately want to do test flights outdoors, but they can begin their research indoors. We needed something smaller that can fly inside and is safe. Indoor lab space is available all year round, the conditions are the same, so it’s much better especially at the initial stages of research.

Jacob: People are doing research in control systems and the control systems can be implemented indoors and outdoors, it does not matter. That is why we started developing what we call the indoor Unmanned Vehicle Systems Lab.

Cameron: We wanted to create a framework for controlling multiple vehicles, the mission development framework. That’s how the Quanser vehicle abstraction layer (VAL) started – a mechanism for developing multi-agent missions and doing high-level vehicle control. We also investigated and developed a number of different vehicles, including quadrotors, Zagi fixed wing, and these were all flown outdoors. Basically due to the reasons Jacob mentioned - environmental factors and the legal restrictions, we started developing the UVS Lab, which is the concept for indoor lab, for doing unmanned vehicle research. Each vehicle is pretty individual in terms how it works. At the higher level, controlling the mission is something that we learned to do through all of the vehicles we investigated and created. Now we can apply what we learned and developed to the newest vehicles that are being released for use by academics.

The newest vehicles we developed were the Qbot, an unmanned ground vehicle and the Qball-X4, a quadrotor UAV designed for flying indoors. The flexibility of these vehicles and the lab setup provide researchers a lot of potential to develop complex missions with even more vehicles, without compromising on safety. With an aerial vehicle like the Qball-X4 there is a higher likelihood of collisions, either with another vehicle or a wall, safety was a big concern. So we decided to build a vehicle that has a built-in cage to protect it and protect the users. We added OptiTrack support for localization purposes, since you no longer have the GPS. As part of our development, the HiQ Aero Data Acquisition card also had to go through many changes. We now have a brand new HiQ that was re-designed from the ground up for these new vehicles. In fact, the Qball-X4 has a completely new HiQ with the on-board Gumstix computer that runs Quarc, Quanser's real-time control software. Throughout all these vehicles the Gumstix has been the main computer on-board. As the Gumstix technology continues to improve, Quanser's unmanned vehicles will feature more powerful on-board computers. While we accomplished a lot in the last five years, Quanser's journey to give the academics a reliable and robust UVS platform has only began.

Quanser's new Qball-X4 quadrotor

Qanser's Leading-Edge UAV Research Project

One of today's hot research areas is the field of cooperative, unmanned autonomous vehicles. We in Quanser want to stay on the leading edge in research and are excited about applying our controls expertise to this opportunity.

The overarching goal of our research project is to permit the seamless integration of commercial-off-the-shelf unmanned vehicles into a fully autonomous, multiagent network. This integration is designed to permit multiagent mission planners to develop scenarios and vehicle objectives without the need for intensive programming or detailed vehicle specific knowledge. Users can treat all vehicles in the network in similar fashion and can focus on specific mission level differences such as different sensors and vehicle capabilities.

Ernest (in the middle) in the CTV studio on May 27, 2008, after giving an interview on Quanser leading-edge UAV research to a TV host Omar Sachedina (left), accompanied by Quanser Engineer, Rajibul Huq (right).

In addition to this work, which is being built to allow for mission designers to go from simulation to hardware as quickly and effortlessly as possible, we are focusing our attention on permitting true autonomous control of these systems. This level of development requires a team approach and we are focussing on accomplishing the research through a close collaboration with Precarn, Defense Research and Development Canada (DRDC), the University of Toronto’s Institute for Aerospace Studies and the U of T’s Joker Hill Ecological Preserve, for example.

Many open questions remain, such as truly representative demonstration mission types or appropriate autonomy levels in those missions. We are always looking for input on these and other aspects of the project, so drop us a line if you have some thoughts.

Quanser Unmanned Ground Vehicle Research Team (from left to right): Don Gardner, Andrew Dawes, Mahyar Fotoohi

In short, we are leveraging our and our partners’ work in developing cooperating fleets of fully autonomous vehicles to be used towards a variety of applications from search and rescue, resource surveying, even investigating climate change by monitoring leave-out patterns in forests. Our vision is to see an easy-use research and mission tool where the humans designing the applications serve more as coaches or managers rather than operators. You can only do this by taking the intelligence from the ground station, where it is currently usually found, and putting it on-board the vehicles where it can do the most good.

Dr. Ernest Earon, Senior R&D Engineer, Quanser

Qbot - Quanser mobile robot, represents another possible unmanned autonomous ground vehicle. Qbot is presented at ACC2008 Conference in Seattle and ASEE 2008 Conference in Pittsburgh. For details, contact sales@quanser.com

Formation Flying For Drones - How Far Has It Progressed?

Defence and Security organizations see unmanned air vehicles, or drones, as a great tool for a variety of surveillance applications. Technology currently being developed will allow drones to be linked into formations and survey vast amount of territory, assist in search and rescue missions and operate in conflict situations while achieving the same level of efficiency as manned aircraft. But before the drone formations can be deployed and operate with minimal human intervention, real-time control schemes must be developed and tested.

In their previous work, Defence Research and Development Canada (DRDC) scientists Dr. C.A. Rabbath and Dr. N. Levin focused on passivity-based formation control, developing the theory for controller design, and verifying it using numerical simulations. Now they have teamed with Dr. Jacob Apkarian, Quanser's founder and Chief Technological Officer, to implement the controller on an actual physical system and perform the experimental validation of the passivity-based formation control concept. At the recent 2011 AIAA Guidance, Navigation and Control Conference they presented a paper outlining the experimental results they achieved.

Using an indoor experimental test-bed consisting of Quanser's Qball-X4 unmanned quadrotor helicopters and Qbot ground vehicles, the researchers created a drone team consisting of followers and a leader. Key tests included autonomous drone formation coordination and mixed mobile robot - drone formation teaming. The results of the experiments indicate that a passivity-based formation control scheme produces cohesive formation motion and can be seamlessly integrated with a commercial off-the-shelf drone autopilot.

Click here to read the full paper titled "Experiments with a Passivity-based Formation Control System for Teams of Small Robotic Drones".

You can also visit Quanser's website to learn more about the Unmanned Vehicle Systems Lab, an indoor platform for teaching and research used in engineering departments worldwide.

UAV Demo Day

As part of our ongoing UAV project, we set a date for a mid-term demonstration including several UAVs and an unmanned ground vehicle (UGV). A large group of people showed up, including DRDC program director Ken Hitchmough and scientist Camille-Allain Rabbath, Hugh Liu, Professor at University of Toronto with several graduate students, and an ample team of Quanser employees. Unfortunately, we picked a day when the weather was exceptionally un-cooperative. Heavy snowfall and high winds had been building up and we caught by unseasonably cold snowy October weather. Nonetheless, our guests braved the weather and we were determined to show them something interesting.

The snow wasn't so bad, but the high winds prevented us from flying the helicopter, and when we tried to fly our fixed-wing Zagi UAVs they had a tough time fighting the winds. We did, however, get the UGV to drive around. The UGV can be driven using a joystick, or in this case it drives autonomously to a given GPS location. Its on-board GPS sensor is used by the on-board controller to steer and drive the UGV to the desired position, in this case a blue tarp.

Had the weather cooperated, we would use the fixed-wing UAVs to fly a search pattern, using on-board cameras to find the blue tarp target. Image processing blocks scan the images in real time and look for the target based on size and color. These UAVs then signal a helicopter, which goes in at a lower altitude to verify the target's position using its on-board camera. Once the target position is verified, the helicopter transmits its location to the UGV so it can drive to the target. All of these tasks are completed autonomously with each vehicle's controller running on-board the vehicle itself. The operators can sit at the base station and track each vehicle's progress, but when everything goes well we just sit back and watch it all happen.

Overall, we gave it our best despite the weather, and it was a great opportunity to "show and tell" our work. The next time we schedule a demo it will be summer - but before that a video from our cold demo day will come.