Blog

2020 in Review

2020 will be remembered as the year where paradigms around much of what we do, and how we do them, were put under serious scrutiny due to COVID-19. In Singapore, the response to the pandemic was a gradual process. As the authorities raced to learn more about the virus and implement the appropriate actions, we were left with a period of great uncertainty. In the early stages, classes could be carried out with added precautions (temperature taking, rigorous sanitization, logging of details for contact tracing etc.) but competitions, which would involve mass gatherings in confined spaces, were left in limbo.

The first competition we were slated to take part in would have been the FIRST Lego League, which was to be held in March, followed by Robocup Singapore Open in March-April. Preparations for both competitions were already under way when the virus hit.

A few days into April, the Circuit Breaker measures were announced (Singapore’s version of a lockdown), making it clear that even if a competition were to take place, it would look very different from a regular year.

For some students, this uncertainty meant a lot. Those who are graduating in 2020 (who have now graduated), were working hard towards these competitions, which could well have been the curtain closer to their journey in competitive robotics at the pre-university level, or even be the last time they would be working so intimately with robots altogether. At schools, the impact was felt greatly as we begin working on Robocup for the next year as soon as the school term closes. Much was already done, with some teams close to completing the tasks and beginning on optimizing the performance of their robots. For our private classes, the sentiments were shared, with the additional financial cost that has already been sunk by our students, now without guarantee that the competitions will be held.

At CORE, we felt these concerns deeply. As competitive roboticists not too long ago, we understood what competitions meant. Due to other circumstances back then, we have, too, been personally deprived of the opportunity to take part in competitions after all the hard work had been done (in the case of WRO 2010, we were informed the day before we were slated to fly, that we would not be able to do so due to the heightened terror threat in the region). Yet it was through these experiences that we understood that the optimal outcome could only be achieved if we had the dynamism to re-evaluate situations and exploit the given circumstances.

On forums all over, many (predominantly pre-university level) robotics educators were scrambling to replace classes previously based on the LEGO Mindstorms EV3 platform. Simulators were suggested, including platforms such as the Virtual Robotics Toolkit and CoderZ. These solutions were not new to us, having been in discussions with partners previously about how scaling robotics education looks like. Back then, we firmly rejected the idea as we felt that there were aspects to working with robots physically that simulators could never replicate, not to mention the poor fidelity of these compute-lite variants.

While COVID did significantly alter the parameters we would base our decisions on, one thing was for sure- we should not compromise on the efficiency of learning for our students where possible. As others were focused on the best way to teach without moving too far from platforms they were already using, we took a step back and looked at the bigger picture, instead posing the question “what media is available to us given these physical restrictions, and what skills skills can we impart at an equal or greater efficiency through these media?”.

Eventually, we decided to take a 3-pronged approach. On the technical front, we would focus on Computer Vision and Electronics. We also took the opportunity to look at softer areas through Forums, such as how robotics is applied in the industry, and how COVID, which had placed automation under the spotlight, has affected trends. The final piece would be competitions, accepting that this piece could very well not materialize.

Technical Modules

Robotics is all about how hardware and software come together as a system. As mentioned, using a simulator would be suboptimal for these purposes. While bringing robots home is a possibility, the robots which could be brought home are limited to those which are small and portable, mostly complete, which do not require external infrastructure to function as intended (in the case of competitive robotics, this could be a play field).

Thankfully, a lot of the sensing and electronics we see in robots are what we already have in our smart devices today. Take your smartphone as an example, which is able to help you localize using sensors such as the GPS module, compass, and your cellular modem (indirectly, by triangulating using signal strength). Under the hood, the algorithms involved include transformation into a common mode, (e.g. longitude and latitude), filtering, and fusion. With all our students owning or being able to loan a device, what was left was to figure out what to teach in this vast possibility space. Most importantly, we were no longer subjected to the limitations imposed by simulation. These were real sensors, real devices, and everyone had their own; no need for sharing.

We settled on Computer Vision for a few reasons. Predominantly, it was something that is becoming increasingly relevant in our daily lives (think Instagram/Snapchat filters, face unlock on laptops and phones, even social distance monitoring drones) and is one of the enablers of what is perhaps going to be the biggest application of robotics in the near future- Advanced Driver Assistance Systems, the precursor to fully autonomous vehicles. More importantly, it allowed for creativity and expression, which might all too easily be pent up as we spend day after day at home during the pandemic.

In schools and for our private classes, we taught computer vision from the ground up, explaining everything from the working principles behind cameras (the pinhole model, the shutter speed-ISO-aperture tradeoff, differences between CCD and CMOS sensors etc.) to the underpinnings of the Haar Cascade (features, adaboost etc.) and getting our students to build everything from filters to games.

While hardware simulations lack robustness thanks to the interplay of forces in complex environments, simulation does lend itself well to basic electronics. This is especially so for hobbyist grade tools, which are simple, well-tested systems where effects like parasitic capacitance and electromagnetic interference are kept in check and do not affect the function of the system to any significant extent.

As such, one other area we could effectively teach online was basic electronics and Arduino with Tinkercad. In fact, using a simulator took away the anxiety of potentially destroying costly components- a common occurrence in the classroom when students are encouraged to deviate from set examples and put together systems of their own.

Due to our focus on Computer Vision, and the shift to competition preparation as schedules were gradually announced, we never really got deep into this, but have definitely seen some quirky projects by our students, made possible by the safeguards that a simulator provides.

Forums

Education serves a varied purpose. Of the lot, industry application is a pragmatic one easily understood sans philosophy. Admittedly, this is something we usually do not focus enough on for a few reasons, including that the content we cover is usually coupled tightly with the demands of specific competitions, which tends to dominate the time we have with our students. With COVID taking competitions out of the limelight, we took the chance to discuss the robotics industry, and in particular, how the robotics landscape has been transformed by the global pandemic.

Financial equations now look very different. Once competing factors of production, the reduction or elimination of human labor from the possibility space means that businesses which did not, or continue not to, invest in touchless automation can now be considered to have disregarded workplace and consumer safety. Where legislation comes into play, these businesses may be forced to cease operations entirely. In a post-COVID world, automation in certain industries may be a necessity, not an alternative, regardless of the financial costs. The (perhaps emotionally driven) response can be quantitatively seen as investment in warehouse robotics was estimated to have surged 57%.

Public perception about robotics has also shifted. While robotics and automation were seen as the great unknown by most before the pandemic, the greater threats now are the virus, as well as other shoppers who act as carriers. There is also a greater understanding about the need for automation, which beats hands-down otherwise not having access to daily essentials such as voltage regulators in a human-staffed organization.

These are but examples of the matters that were discussed during these sessions- softer aspects we usually do not have the chance to discuss, yet are fundamentally important as we consider the education of our kids holistically, understanding that the work our kids eventually do will exist in a greater societal context that is the end goal, not automation in and of itself.

Competitions

2020 started with preparations for competitions, as most years do, being ~3 months out from Robocup Singapore Open. We kept working even as things became increasingly uncertain. On one hand, this was to ensure that we were prepared in the case that COVID blows over quickly. More fundamentally, we know that the value in competitive robotics is in the months of preparation, not the minutes of competition.

When Home Based Learning was implemented in schools, robots and tools were brought home and rotated among team members when possible, with consultations done over Zoom. Suboptimal, but we did the best we could, giving our kids a goal to work towards. Without this, I feel we could have lost a lot of engagement due to the lack of a tangible goal. This was something we could not afford as culture is something we have fought hard to build over the last few years.

In March, things started moving. The Robocup federation announced that the international finals, slated to be held in Bordeaux, would be postponed to 2021. Perhaps due to the need to uphold the prestige that comes from the standards associated with the competition, it was decided that there would be no official competition. Rather, each league in Robocup Junior announced their own plans, mostly involving poster presentations.

The league which came closest to replicating the real event was CoSpace (also known as Rescue Simulation), which did not come as a surprise as the regular event involved a simulated component as well- perfect for the situation we are in. The International Cospace Online Challenge was to be an event that was purely virtual (robot-less). Apart from video presentations and interviews, competitors had to go through a preliminary round, where they would attempt the challenge as a single agent. The top scoring participants were then invited to take part in the finals, which would involve a 1 vs 1 match. Rather than focusing on the competitive aspect, where online judging posed many problems yet to be resolved, the competition was held in a “friendlies” format, with awards such as Best Presentation and Best Strategy given out.

Against what may have seemed like my stance (against simulation), I got the support of teachers at schools and some of my private students to give it a shot. Oftentimes, hypocrisy is prematurely called before obtaining an understanding of nuance. I intend to write another piece reflecting on CoSpace specifically as this piece would get too lengthy otherwise. To paint in broad strokes, CoSpace is more of a Robotics Programming competition than a Robotics competition. By using a standard (simulated) hardware platform, it ditches the pretense of being one, and ensures fairness by subjecting all participants to the same idiosyncrasies of simulation. Without any significant interactions between robot and environment, which is typically where inaccuracies show, there is little that could unfairly disadvantage certain strategies.

The experimental nature of the event meant that instead of being a week long affair, proceedings were dragged out over weeks, concluding with an Award Ceremony in the middle of August. While this resulted in some frustration at times, empathy, patience and understanding are going to be key amidst the uncertainty which will undoubtedly persist as we continue to learn and experiment to develop solutions which can exist anywhere on the virtual-physical continuum on demand.

In June, it was finally announced that FLL would be going virtual. This time, the Robot Game component was to be scrapped, with the presentation component remaining (done over Zoom, of course). Just based on the possibility of technical execution, I felt that the Robot Game could have remained, even if the judging procedure could have been tedious for both teams and referees. After all, teams typically purchase the Playing Field and season-specific parts required to assemble the mission elements. It is difficult to alter dimensions significantly with the LEGO components that form the environment, making it difficult for teams to cheat.

However, I do agree with the decision to conduct the competition in the format it was eventually done in. Given that this was fairly early on, COVID was wrecking havoc on all of our lives. The outcome of holding the Robot Game virtually would be that teams would be evaluated based on how prepared they were at the point at which the Circuit Breaker measures were announced (penalizing teams which were planning to work harder closer to the originally planned competition date), and how well their circumstance happened to suit the restrictions imposed by COVID (schools with many teams sharing one Playing Field would have a logistical nightmare in terms of testing).

At CORE, we met online to prepare for the presentation component based on the work that was already done at that point. While it was most certainly incomplete, we could not risk conducting physical sessions and subjecting our kids and the people around us to the whims and fancies of a deadly virus we did not know enough about. As always, the presentation component was the perfect opportunity to consolidate learning, further enhanced by the fact that this was now our sole focus.

In July, the Robocup Asia Pacific committee announced that the competition would be held virtually, having learnt from the experiences garnered from the international events. This time, the CoSpace category was no longer going to just be a friendly. Through a preliminary selection stage followed by group and knockout stages, a great number of participants could be admitted and filtered through (relatively) fairly. Our kids also warmed up to the category and took this opportunity with much more certainty than at iCooL.

I was also part of the Robocup Asia Pacific Organizing Committee, judging presentations for the RCJ Soccer Category where teams were evaluated based on a video, poster, and paper. Truth to be told, it felt like a physical competition transplanted online. With Technical Challenges already a part of physical events, I felt that much more could have been done to evaluate the capabilities of the teams even in the absence of the adversarial element. After all, Soccer is hard enough such that the teams which can complete the challenge would be able to do so with a clear margin, eliminating the really close calls we usually see in WRO/FLL.

At the beginning of September, WRO, like Robocup, announced that there would not be an official event. In place of it, the Canadian national committee (original host country of WRO 2020) announced that they would hold an unofficial event- WRO 2020-X. Having been caught up with Robocup preparation up to this point, I floated the idea of taking part to my students, without any serious intention. The enormity of the task was apparent, with none of them taking it up.

However, I did want to go behind the scenes and gain valuable experience running a virtual competition, and to understand how teams were coping and preparing during the pandemic. I registered to be, and was subsequently selected as a judge for the surprise category. In November, the event was successfully run. For a first time attempt, the event was executed excellently. I was pleased to see that participants were much more relaxed than what I was used to. This allowed for emotional investment in learning (as their expressions made clear), something competitions should achieve, but may find it hard to given the environmental pressure on students to excel in a typical setting rich with pomp.

The end of November was also when Robocup Singapore Open was held. By this time, our students were truly tired out from the demands that COVID put on them on throughout the different aspects of life. Academic calendars were crammed, and we were coming off 2 competitions in less than half a year. We left this competition optional for them, but encouraged those who could afford the time to have a shot on their own (no official sessions at schools and minimal Zoom meetings for our private students). The format adopted was similar to what was done at RCAP, with the quickfire nature of these events meaning that there was little time for evaluation and re-development anyway.

* While the National Robotics Competition (NRC) was also held, we did not take part in it for a few reasons. Among the 3 categories, Coder Z was not considered for reasons that should be clear, the DJI Robomaster was too hefty an investment for a platform which is not well established, and WRO Open was a little too rushed given that we could not even guarantee physical access and training.

Results

Robocup Singapore Open
– CoSpace Grand Prix U19 3rd, 4th
– CoSpace Rescue FirstSteps U12 1st, 2nd
– CoSpace Rescue FirstSteps U19 1st, 2nd, 3rd, 4th, 5th, 6th
– CoSpace Rescue U19 1st, 2nd, 3rd

Virtual Robocup Asia Pacific
– CoSpace Grand Prix U19 1st, 2nd, 3rd
– CoSpace Grand Prix U19 Best Strategy
– CoSpace Grand Prix U19 Best Presentation
– CoSpace Rescue FirstSteps U12 3rd
– CoSpace Rescue FirstSteps U19 1st, 2nd, 3rd, 4th
– CoSpace Rescue FirstSteps U19 Best Strategy
– CoSpace Rescue FirstSteps U19 Best Presentation, Best Presentation (Joint)
– CoSpace Rescue FirstSteps U19 Best Team Description Paper
– CoSpace Rescue U19 Best Presentation
– Most Popular Video 1st
– Most Educational Value 1st, 2nd, 3rd
– Outstanding Performance Award (Institution) for Raffles Institution
– Outstanding Performance Award (Institution) for Raffles Girls’ School
– RCAP Community Engagement Award (Kenneth)

International CoSpace Online Challenge
– CoSpace Grand Prix U19 Best Presentation Award 1st, 2nd, 3rd
– CoSpace Rescue U12 Best Presentation Award 1st
– Cospace Rescue FirstSteps U19 Best Presentation Award 1st, 2nd, 3rd, 3rd (Joint)
– CoSpace Rescue FirstSteps U19 Most Creative Presentation Award
– CoSpace Rescue FirstSteps U19 Best Strategy Award
– Community Awareness 1st (Session 4), 3rd (Session 1)
– Most Educational Value 1st (Overall), 1st (Session 2), 1st (Session 4)
– Most Popular Video 1st (Overall), 1st (Session 2)
– People’s Choice 1st (Session 3)

FIRST LEGO League Singapore
– Robot Design Award (Mechanical Design Awards) 1st

Have we achieved what we set out to? Competition results and my personal evaluation of the growth of our students suggest so. That said, there is still much to learn. Perhaps the most important things that COVID has taught us would be to be more understanding of each individual student and their circumstances, and that positivity can make a difference. Ultimately, as we have always believed, results from competitions seldom adequately reflect the value they add. Between fortuitous victories and minor missteps leading to disastrous outcomes, this list serves as a nice validation of the work we do everyday, sometimes with less visible outcomes.

What next?

As much as COVID-19 has been a global catastrophe, it allowed CORE to show off our dynamism and astute decision making in the face of massive uncertainties. Personally, I feel that I have grown from this experience, both as an educator, and in my professional capacity.

While we may hope that next year will be better, that is likely what it will remain as- hope. The curveballs that we have been thrown should remind to be on our toes.

2020 has been a worthy challenger, and I feel that CORE has responded brilliantly. We took things as far as we could, grabbed the opportunities that came our way, and provided our kids with a learning experience that would rival our offerings in any other year, albeit with a shifted focus away from the hardware aspects of robotics.

Ultimately as the Centre Of Robotics Excellence, I do hope to return to making things. But in the absence of that, we will keep doing the things we have done well in 2020, and improve on this foundation in 2021, as improbable as that may seem given the results we have achieved.

The New Year is yet another arbitrary marker in time. Yet that does not mean it lacks implications as so many of our processes are tied to the passage of 365 day blocks. Already, we have been reflecting and making plans. There is so much to be done, and we look forward to be taking everyday on, learning and growing with our kids.

Bring it on, 2021.