Answers to your questions about the CSA's rovers
On , a team of engineers from the Canadian Space Agency (CSA) hosted a Reddit "Ask Me Anything" (AMA) to answer questions about the CSA's rovers.
A variety of topics were covered, including the Lunar Exploration Analogue Deployment conducted by the CSA between and . The mission consisted of a series of field tests to replicate scenarios of a sample-return mission, using the Juno rover.
Several people participated in the Reddit AMA. If you were unable to participate, here are the answers from our experts.
- Pierre Allard (PA): Robotics engineer, specialized in ground control systems
- Chantelle Dubois (CD): Computer engineer, former intern from the CSA trainee program
- Sébastien Gemme (SG): Robotics engineer, specialized in software and computer vision
- David Gingras (DG): Robotics engineer, specialized in guidance, navigation and control
- Natasha Jackson (NJ): Robotics engineer, specialized in mechanics
- Tom Lamarche (TL): Robotics engineer, specialized in electronics
- Martin Picard (MP): Senior robotics engineer
Questions from participants on Reddit and answers from our experts
Technical specifications on rovers
What software is used in the development of the rover and the mission?
The software on the rover will make use of many libraries. We are currently looking in using the core Flight System (cFS) platform developed by NASA for the management of the various apps used to receive commands, collect and transmit telemetry, manage telecommunications, control actuators, maintain localization estimate, perform imagery acquisition (2D and 3D and control payloads).
For prototyping, we are making use of the Robot Operating System (ROS) since it provides a lot of already built features that makes prototyping fast and easy. We use Apogy for the ground control system when testing our Earth prototypes in simulations on Earth. – PA
Can I have some basic stats of the rover? Dimensions, weight, instruments on it, etc. If not, can I at least have a nice HD picture?
The Juno rover in its configuration for the Lunar Exploration Analogue Deployment (LEAD) mission simulation has total mass of 445 kg, which includes all mounted equipment. The rover would tightly fit in a rectangular box of 1.6 m × 1.5 m × 2.1 m (W×L×H). The rover is equipped with various instruments and sensors. There is a robotic arm that includes a scoop-rake-gripper tool to collect soil or rock samples and to transfer them into an on-board sample canister. The perception sensor suite includes several vision systems such as a laser camera (a.k.a. lidar), a stereo camera, a pan-tilt-zoom camera, 3 wide-angle drive cameras as well as 2 cameras dedicated to providing imagery of the robot arm's workspace. Here is a photo of the rover and some descriptions about the on-board equipment. – DG
Will there be publications and reports about the project and missions? Some questions in particular about the analogue missions :
- Can and will the Apogy software be adapted for use in space and on the Moon?
- Did you also work with ESA's groups doing similar activities, e.g. the Analogue-1 missions involving ESA's European Astronaut Centre (EAC), European Space Research and Technology Centre (ESTEC) and even ESA astronauts on the International Space Station, or even the robotic HCI-related work being done at both EAC and ESTEC?
Two papers were presented very recently at the iSAIRAS conference. One outlines the overall LEAD deployment and the associated results, while the second one focuses on describing the Sample Handling Subsystem. Feel free to have a look at them to get the details. A lot of (good :) work went into this project, and we did put a fair effort into trying to condense all we did and learned into those articles :
- Lunar Exploration Analogue Deployment (LEAD): Overview of the - Robotic Sample Return Mission Simulations (PDF 1 140 KB)
- Sample Handling Subsystem (Shs): Concept Demonstration for a Lunar Sample Return Mission (PDF 4 610 KB)
Apogy could indeed be used in a space/Moon context. The Apogy framework has been developed explicitly to target multiple space missions. Different environments, and modules to model the Moon and its surface have been recently added. – PA
We have not been involved directly in the activities you are referring to, but we are aware of them and the CSA has regular interactions with partners in the ESA family. The LEAD deployment was in fact a direct collaboration with ESA and their HOPE activity. Those are certainly interesting topics of exchange for our group. – TL
Hello. What are your hardware specs? Computer. Are your systems networked and can they theoretically be hacked? Do you use machine learning? Thank you.
On flight projects, we use ARM and Power PC based architectures with field programmable gate array (FPGA). Those processors have the advantage of providing a good ratio of processing power vs. watts. Most of the platforms we use are fully radiation hardened but sometimes, we need to trade radiation hardness for processing power and use processors that are resistant to radiation, meaning they would not easily get damaged by radiation, but that may from time to time reset due to upsets, when a high-energy particle hits the board for instance and causes a short in the electronics.
As for networking: our systems are reachable from the ground, from multiple sites, using a secure communication link with different levels of firewalls and they could theoretically be hacked, so, IT security is taken very seriously on flight programs. As for machine learning, we are starting to look into machine learning applications for both the ground and the on-orbit segments. The on-orbit segment is more challenging because of the limited selection of hardware available, i.e. we cannot run a high-end Graphics processing unit (GPU) on a spacecraft because only one GPU we use on Earth needs more power than the total amount of power available on the spacecraft. – SG
Can the rover automatically detect obstacles? If so, what algorithms are in use? Can it avoid obstacles by itself or does it need manual operations?
Yes, we are using a suite of sensors, providing a depth perception, to detect obstacles, namely LiDARs and stereo cameras. For LiDARs, an algorithm analyzes the 3D point cloud to assess obstacles and produce a traversability map. For stereo cameras, we can work on the point cloud or on the disparity map. This information is then used by the rover autonomy software for path planning purposes or unexpected obstacle detection. For instance, in the case of an unexpected obstacle, the rover will stop and wait for an input from the operator when in manual mode. On the other hand, when in autonomous mode, the rover would stop and plan another path, taking into account this new obstacle. – SG
Yes, Juno has on-board a software that enables autonomous navigation capabilities. The operators specify a commanded destination to reach that could fall beyond the rover's sensing horizon. This software relies on dense 3D point clouds acquired using the lidar to sense the rover surroundings. A terrain assessment is performed from the point clouds in which risky areas are tagged as unsafe and benign regions as safe. Then a collision-free path not longer than 6 m is planned. The rover would follow that safe path relying on its on-board path-tracker control system to reach an intermediate goal. Once arrived at this intermediate goal, another lidar scan is taken, and the planning and motion loop repeats until the rover reaches the commanded destination. In theory, under the autonomous navigation mode (Autonav), the rover usually does not have to react to unexpected hazards since it carefully analyzes and plans its motion around the obstacles. But in the real world, the vision sensors are not perfect and the algorithms under the hood rely on assumptions. So it happens sometimes that the rover misanalyses the terrain and has to automatically stop in the middle of a path because there is an unexpected hazard on its way. That's thanks to the stereo camera, which continuously assesses the traversability of the terrain in front of the rover at a frequency of 5 Hz. When an obstacle is detected, the rover automatically applies a panic brake and restarts its Autonav loop. So then it would take a lidar scan and "see" the obstacle. It would then plan a path around it and resume the navigation to its target destination defined by the operators. – DG
How much of a role does machine learning currently play in your autonomous navigation system?
Currently, machine learning is not being used for rover navigation; we are using traditional methods. Those methods currently support the concept of operations we need to implement. However, machine learning is of interest to the CSA, and we are looking at eventually integrating it into our autonomous navigation ecosystem. For instance, we are currently prototyping a system that is using deep learning to perform terrain assessment. – SG
What type of suspension does it use? Thanks for the initiative!
This rover has a passive suspension that allows the drive units on both sides ("bogies") to pivot around a central axis. This guarantees that all four wheels are continuously in contact with the ground, to maximize traction and stability. It also has the effect of averaging the chassis tilt as the wheels drive up and down obstacles. Since this suspension architecture is passive (e.g. no actuators, no active control), and does not require any springs or dampers, it keeps the design as simple as possible, which in turn makes this approach very reliable and gives it a better path to flight. – TL
What role does Juno play in the overarching Artemis program?
Juno is a terrestrial prototype rover that has been used in many different terrestrial analogue missions and tests. Canada could eventually use technology developed for Juno and bring up to a flight rover, for example the technology developed for its drivetrain (wheel, suspension, motors). – MP
What are some of the mechanical challenges your team has had to overcome with the development of the rover? Such as testing different material strengths and designs for construction?
As with all space missions, the constraints on mass is one that affects most of the design. All structures must be light while being robust enough to endure the harsh launch environment (vibrations). The structural materials to be used will most likely be aluminium alloys, titanium and possible carbon fibers.
Since the launch is by far the most difficult environment, the rover mechanical design will have no problem surviving Earth's gravity during tests. The exception here are the motors and gearboxes which will be sized for the Moon's gravity, which will reduce the rover performance in slopes in Earth's gravity.
The thermal environment on the Moon is very challenging, especially when trying to survive the night (which is close to 2 weeks long!). Night survival will involve maintaining parts of the rover (batteries for example) at higher temperature than the frigid Moon night, which will require active heating. Since the batteries have limited capacity, the volume that requires heating (the Warm Electronics Box) needs to be as small as possible and well insulated. The same design must also allow excess heat to be rejected when the rover is operating during the day in full sunlight. It's a bit like trying to design a winter coat that keeps you warm in the coldest of winter and yet keep you cool in the middle of the blistering heat of a midsummer day. – PA
Are the Juno drive cameras standard IP PTZ cameras? If so, how do you deal with the associated delay of around 4 seconds on the RTSP stream?
We do in fact use IP cameras, not streamed but polled at low rates, such a 1 image per second in order to save on bandwidth. There are no perceptible delays when viewing the image. We do, however, artificially inject a communication delay, around 5 seconds, to simulate the communication delay between Earth and the Moon. – SG
Do you guys use ROS on Juno? By any chance do you use Open MCT?
We use ROS on Juno (which I'd like to remind everyone is a prototype rover, so not a flight model!). Instead of Open MCT we use Apogy, which our colleague, Pierre Allard, developed. – CD
We have been using ROS and Apogy for many years now (see the CSA website on the Mars Sample Return Mission Simulation and the Lunar Exploration Analogue Mission to integrate sensors and robotics arms to rovers. – PA
What type of communication do you use in Juno? Wi-Fi only?
We locally use Wi-Fi to communicate with the Juno rover. However, when the rover is deployed in a remote area, we use an internet satellite communication link or cellular network to communicate with the Wi-Fi infrastructure at the remote site, to which the Juno rover is connected. – SG
Nice rover! Many similarities with our Juno. For instance, it uses the same passive suspension architecture. – TL
Comment from participant: Thanks! I just finished coding the differential drive and I also developed a cloud control system for it. As long as the rover has internet I'm able to teleoperate it from anywhere.
Are those commercial security cameras on the rover?
The Juno rover is a terrestrial prototype that we use to test concepts of operation. So yes, when we needed to deploy cameras on the rover to get images, we selected COTS security cameras. These are usually very simple to integrate, but they have no "path to flight." For more advanced applications such as 3D perception, we use high-quality scientific cameras. – DG
Is Moon dust a challenge for a Moon rover? If so, what measures have been taken to mitigate those challenges?
YES! It's a big challenge. Moon dust is highly abrasive and very hard on moving parts. It also tends to become electrostatically charged and can stick to surfaces like solar panels, radiators, and sensor lenses.
For mitigation we have to pay special attention to hardware seals (as well as tortuous paths for the junction between the parts), size our systems to operate after they are degraded by dust (e.g. oversize solar panels at the start of a mission to deal with power reductions over time, plan for increased drag in the locomotion systems), and we have done a lot of work on wheel development to operate over long periods on the lunar surface. We also do extended life testing of items like the drivetrain in dusty environments to identify any potential weak points in the design.
Active repulsion systems have also been tested for removing dust on the radiators or solar arrays. –NJ/MP
When can we expect the rover to land on the Moon?
The CSA is actively planning to put a small rover (~ 30 kg) on the surface of the Moon under its current Lunar Exploration Accelerator Program (LEAP) by . – MP
What were some of the mechanical challenges in compensating for the lower gravity environment of the Moon?
Actually the Moon's gravity is of great help in terms of mobility at the surface of the Moon. Given that the gravity is one sixth of the gravity on Earth, that results in a reduced need in energy to move the rover around. This is a very good advantage. Another advantage is that the rover structure also supports a reduced gravity, however it is difficult to test on Earth an optimized rover for the Moon. – MP
What questions are you hoping to answer with this rover and the lunar sample return mission in general?
The rover would be used to bring back samples to an ascent vehicle that will then bring those samples to Earth. Scientists will then look at the samples, similarly to what was done for the Apollo missions. The goal is to understand the formation of the Moon and also to identify potential resources that could be used to settle a human base on the Moon. – MP
Canadian Space Agency in general
How closely do you work with other space agencies (e.g. the National Aeronautics and Space Administration (NASA), the European Space Agency (ESA) for your work, and how has COVID-19 affected international collaboration?
We do have a number of projects with other international partners around the world: NASA, ESA, the Japan Aerospace Exploration Agency (JAXA), etc. With COVID-19 we had to reduce the travel and face-to-face meetings to none. This involves working remotely via the internet and also having to attend meetings sometimes quite early in the morning or later at night, but we keep working together and exchange on the situation. – MP
Does Canada have its own space agency/rocket to get into space, or are you partnering with another country? If so, who?
Canada has a space agency. ;) It is called the CSA. We partner with international entities to coordinate and negotiate launching our spacecraft. For example, the RADARSAT Constellation Mission was launched with SpaceX in the United States (USA). We currently do not launch from Canada. – CD
Not yet! SpaceQ, a Canadian media outlet covering the space industry, keeps a good record of the launch development efforts being made in Canada by various independent entities. – NJ
What is the best way for Canada to advance its interests in space without a launch capability? We've seen in the last 4 years that nothing about our traditional partner can be counted on as solidly reliable in the long term. Is it time for Canada to invest in its own launch facility, or provide support and partnership in an existing initiative?
Transport Canada is responsible for rocket launch activity in Canada. For now, Canada has made a decision to take advantage of the existing national and commercial launch capabilities around the world. Although the Government of Canada has not invested in a national launch facility, we are encouraged to see the commercial interest in such a development.
Launch as an open market? Sounds like a great reason to develop the capability here! Can I ask a follow-up? Would any and all development of a hypothetical Canadian launch facility need to be on the East Coast (like the proposed Maritime Launch Services facility)? Or for our purposes, would a site on the western Hudson Bay or the West Coast be feasible/safe?
Launch is actually becoming a very open market. Those responsible for a mission buy a ride on a rocket, dealing directly with the launch providers (e.g. SpaceX) for most Earth-orbiting platforms now. Like chartering a plane... ish.
The Moon is starting to open up this way as well with the US CLPS program.
Beyond, well, we work with many partners and everyone is working towards common exploration goals.
A homegrown launch capability would be great, and some companies are working towards it. It's becoming a commercial venture first. – NJ
I suspect that it's desirable to launch rockets over the ocean and not the Canadian landmass, but I am definitely not an expert on this specific topic, so that's just my general take on why the East Coast seems more desirable. I invite my peers to provide input if they have any further insight. – CD
Yeah, the East Coast is certainly the low-hanging fruit in terms of starting a Canadian launch facility. It's well outside the scope of our group, but the other two sites would most likely be very limited in the launch trajectories that would be acceptable. For the West Coast, you can look up the limitations for Vandenberg AFB – it would probably be a similar situation but with less flexibility on orbit inclinations. Launching eastward is also helpful because you benefit from Earth's spin to provide extra speed, which can be leveraged over the ocean on the East Coast. – NJ
Our experts and space careers
I have wanted to work for the CSA since I started my undergrad, but after not hearing back on many applications I've kind of given up. Is it easier to transfer into the CSA from another agency, or how do most people go about getting their jobs there?
The early career job search can be challenging. I had my fair share of applications I never heard back from as well. I would highly recommend that you keep applying, and apply for everything you're interested in and qualified for. We post many student positions every year that are evaluated by different hiring managers, looking for very different skill sets. You may still find the right match.
Post-graduation, there are opportunities to enter the CSA directly or to transfer from another government department. Most job postings are public, and everyone (at another government agency or not) applies the same way. As Chantelle mentioned in a different question, it's a great idea to create an account on the Government of Canada job website. You can add filters for daily notifications (filters like "CSA," "Engineer").
Other than that, just keep building your experience – go to conferences, do projects, get jobs that will let you develop transferrable skills. You can keep building towards a job at the CSA or in the space industry in many ways.
Another thing to think about – the CSA implements most of its programs by contracting Canadian companies to do technology and hardware developments. You can work on CSA programs, usually in a more hands-on technical role, through these companies. Experience in the Canadian space industry would also be great on your résumé when you found the right opportunity at the CSA. – NJ
How have you been able to learn and become experts in space-focused robotics, if most standard engineering, computing and robotics education doesn't cover this kind of specialization? Are there any resources that you'd recommend? And how can a student get involved as an intern, similar to Chantelle's internship in the trainee program? Does this program still exist?
On the point about how to get involved as an intern, the CSA posts co-op opportunities directly to university co-op programs in Canada and occasionally hires through FSWEP. You can ask your co-op program office if they can reach out to the CSA to establish a relationship if one doesn't exist already, and for FSWEP you can add your profile anytime although it's a bit random on which government entity will receive your application.
For the Engineer Development Program, the posting opens every few years and is posted on the Government of Canada job website. I recommend creating an account and adding filters for daily notifications (filters like "CSA," "Engineer"). This program is meant for engineers who have recently graduated and are still at the junior level.
On how to get space-focused training because standard engineering programs don't offer it – joining or creating design teams for nanosatellites or robotics is a good way. Otherwise, a lot of junior engineers join the CSA with limited space experience and learn the specialized knowledge and skills as they go along. I think it's easier to just be a really good computer/mechanical/electrical engineer, and then know how to adapt those domains to space. You can also try to choose space-related capstone projects or pursue a master's focusing on space applications. I didn't study space computer engineering, just regular computer engineering. But I was part of a nanosatellite team, so that helped me understand what sort of implications space applications have on computer engineering work. – CD
Who are the people in the robotics mission control centre (or ground team), and how should one follow a career pattern to join the team? Or more specifically, what would be the best to focus on after I complete my aerospace engineering degree if I want to work in mission control?
The people you see in the classic mission control room shot in every space movie are a combination of specialists, probably engineers (the combination being dependent on the actual mission happening). There's no tried and true pattern or path to get into those roles. Some of those individuals will be people who work for the company that built the robot/spacecraft they're controlling (the contractors) who are there because they know the system well and know how to operate it. Others are trained specifically to be flight operators or flight dynamicists. I suspect those roles are easier to get into after being in a systems engineering type role, or if you are in some way involved in the building of the spacecraft. If you look up Kristen Facciol on Twitter, she is one of our mission controllers and she posts a lot about her job! She went to Houston to train for 2+ years in order to control Canadarm2. I believe she previously worked for MacDonald, Dettwiler and Associates Ltd. as a systems engineer. – CD
What's your engineering background? I'm a junior manufacturing engineer with a mechanical engineering degree making prototype precision gears for big EV companies, but I'd love to one day work in the aerospace industry!
My background is software engineering and a master's specialized in computer vision. – SG
I studied computer engineering (BSc) at the University of Manitoba. My work at the CSA involves robotics software, ground segment software, and systems engineering. I've also dabbled a bit in (space) policy before. – CD
I have a technical degree in industrial electronics, then a B.Eng. in electrical engineering. I got really into robotics at the CSA, and eventually completed a master's degree part time while working full time at the CSA, in electrical engineering, focusing on lidar sensing for rover navigation. – TL
I have a B.Eng. in aerospace engineering from Carleton University, with a specialty in space systems design. I worked in testing and design engineering in the aviation industry before coming to the CSA.
Sounds like you're getting some cool experience. I think everyone benefits when there can be knowledge transfer between industries. Maybe we'll see you in aerospace some day. – NJ
I have a B.Eng. in mechanical engineering and a master's degree in electrical engineering. My master's thesis was about rover perception and motion planning. – DG
So, uh, you guys looking for a welder? Haha just asking for a friend... yeahhh
You know, we do have a prototype shop... the CSA employs technicians to work in our machine shop, in electronics fabrication, and to do rover maintenance and other roles. We're not all engineers and scientists here (we have finance, admin, communications, and many other roles as well for that matter). Technician postings are somewhat rare – it's a smaller group, but you will also find them on the Government of Canada jobs page and can set up notifications for the EG group. – NJ
What's the cafeteria food like at the CSA? Is there a secret phrase you give to the kitchen staff for the "good stuff"?
I personally think it's really good! And affordable. I used to get coffee from there a lot. If you are really nice to all the staff, they recognize you and will probably give you the good stuff ;) I appreciated that they always had at least one vegetarian option on the menu every day. I miss eating with my colleagues there. – CD
So our headquarters is in Longueuil, Quebec (QC). Quebec is known for its maple syrup, and there is a "sugar shack" day when you can get traditional sugar shack breakfast drenched in the stuff. – NJ
Very nice cafeteria with many choices and great thematic days. – MP
I agree that food is (or was before COVID lockdown!) quite good and affordable. There was, however, a dark age many years ago where quality got relatively mediocre, but fortunately this is a thing of the past. I don't think there's a secret trick to get the "good stuff," other than being nice to the cafeteria staff, but if anyone knows something I don't, please share! – TL
Considering all the problems and challenges you had to overcome to achieve such great lengths for all the rovers, what would you say is the most fun you had on the project so far and yet to come?
For me personally, the joy of seeing something I wrote (software) do what I want it to do; learning from my colleagues; sharing successes together. That's the most fun. I think we all like our jobs very much. Although I have a story about Sébastien: one time we were in the Mars Yard testing the rover, and out of nowhere he asked me, "Do you know the word trouser?" He had been helping his kids with their English homework and he had never encountered this word before, and for some reason this random question killed me and I probably laughed for a solid 5 minutes. – CD
Seeing all the subsystems on which we worked for quite a long time converging together into a complete analogue mission with our international partners that has been very successful was really great! – MP
For me, the challenges and countless hours trying to make things converge from "is there a way to do this?" to something that actually works is what in the end brings that inrush of satisfaction, when you finally get there: someone clicks a few buttons, the rover does its thing, all of a sudden making it look so easy (knowing very well that what is behind is not). – TL
- Date modified: