Beyond Rule Engines: Advanced Lighting Control for Home, Office and Industry with ubisys Automations

Simple Rule Engines…

You might have used simple rule engines to automate your home à la “if this then that” (IFTTT) or the automation features built into popular home control apps, such as Apple Home or Google Home. These are super convenient for quickly setting up automation rules. Simply select triggers and actions and the hub will execute the rules you have configured. For example, occupancy sensors detecting human presence could turn on specific lights in related areas. Setting up such an automation is done simply by picking items from lists of devices, traits, primary and ancillary conditions.

… and their limitations

What if you need more sophisticated logic? Say you would like to turn on the light, but only if there is not enough daylight; because what is the point in turning on lights when the room is already illuminated by sunlight? So, you add another sensor (an illuminance sensor) and another condition: Turn on the light when occupancy is detected (i.e. someone is moving around in that area), but only if also the light level is below a certain threshold. A logical??“and” condition. Simple enough.

Now, what if the lights are dimmable, and we want to smoothly fade in some artificial light? Just enough to reach a setpoint for the target light level when combined with light from other sources, e.g. daylight. After all, harsh on/off lighting automations are more than annoying and we are talking about?smart?buildings, not an el cheapo PIR sensor wired to a relay in your front yard. To realize such a daylight harvesting feature, you could maybe try and add more rules, e.g. 10% light output when illuminance is more than 400 lux and less than 450 lux; 20% between 300 lux and 400 lux, etc. pp. If you were at all able to enter a rule with such a range constraint in the first place.

There is another small thing we missed: That light would remain on forever… Guess what, we would end up adding another rule: “When there is no more occupancy turn off the same lights”. And more features, more rules… you get the point. A recipe for catastrophe. A nightmare for maintenance.

When the rubber meets the road

Now, let's assume you somehow managed to get all the rules in place and deploy this kind of logic in an office building. It won’t take too long before the support phone rings: “I have turned on the light, but it keeps turning off all the time! It drives me nuts!”. With a calming voice, technical support would respond: “Sure, that is the intended behavior. It is fully automated now. Great, huh?”. Even through the phone, you can literally feel the pulse of the person at the far end rise to a life-threatening 180+ bpm. Another classic: “I am sitting here, and the light always turns off after a minute unless I keep moving!”. That poor support person thinks: “What do you expect? You should not be sitting around all day on your fat a** [beeeeep].” But thanks to numerous customer relationship trainings and courses he manages to reply: “Um, no problem, sir, we might be able to extend the time for you.” Now, multiply this by the number of users, exponentiate by the number of occurrences… and watch your support hotline collapse.?

You did not factor in the human component. Maybe they did not like the threshold that was set for them once upon a time, perhaps under different lighting conditions. Potentially, the interior has changed from a dark gray carpet to a bright white floor after the system was installed. Reflections, refractions - all different now. Most of the time the preset light levels might be perfect, just sometimes Ms. X prefers this setting while Mr. Y prefers another. Occasionally, they need more light for a specific task. Conversely, sometimes they need dim light when there is a meeting in a conference room with a presentation being held.

When someone is using a light switch or nowadays an app or voice control to dim the light to a specific level, they would expect to suspend and overrule the automation. But for how long? And how could you even achieve such a behavior with a simple rule engine? You would need state information, contextual prioritization, and other building blocks not found in a typical rule engine.

To the rescue: Scripts! Or not?

So how to overcome these limitations, in particular when problem complexity grows exponentially as users and customers demand more and more features? Your best bet is an environment that lets you run programs or scripts. That would allow you to implement even the most exotic customer requirements. Hence, it is no surprise Google just launched a scripting feature for Google Home Routines last week.

However, anything to do with code is for power users. And for power users, only. Even if they just need “basic coding skills” this is where the party is over for 99% of mainstream customers who just want to use a system. They are interested in utilizing a solution. Not in building one. Even if they had the skills, they might not have the time and patience to develop a full-blown lighting control solution, for example. Smart systems are meant to save their users time and money, right? Not engage them in developing building automation scripts. That is something for hobbyists to spend their spare time on while having fun with a Raspberry Pi or for professionals offering development services and charging their customers for it.

Some vendors solve the problem by fully hardwiring the logic of the control system, i.e. making the feature set static and leaving the user with some limited configuration options. This makes it simple. But also limiting. Either it works for you, or it does not. No way of extension or customization unless the vendor updates the system. There must be a better way.

Come on, it is 2023. AI can certainly do it all for me, can’t it?

Wouldn’t it be wonderful to simply ask AI to do it all for you? Let’s imagine a fictional agent, Alfred. A super-sized cousin of Siri, Alexa, ChatGPT, Bard and Bixby, an open-source evolution of LLaMA, Alpaca and other animals. After years of research and training and tons of investments, Alfred – you might also call him Al ("A" and lowercase "L", potential confusion with "AI" in sans-serif fonts is fully intentional) – is finally ready to take on the challenge.

Let me help you draft the prompt, which seems to be the only work left for us humans:

“Al, I would like you to generate code for a lighting control system. Target the JavaScript automation engine of the ubisys G1 for deployment. Expect about 100 to 150 Zigbee lights and sensors to be connected to a single G1. Typically, one G1 per floor. Roughly 10,000 light points in the project. The system shall support human centric lighting, daylight harvesting as well as schedule and occupancy-based control for a typical office environment. Create different schedules for weekdays and the weekend. Also, I would like different settings for office hours vs. night hours. I will fine-tune the specifics later. In addition, manual control via wireless energy-harvesting switches shall be possible. Specifically, dimming, selecting scenes and pausing/resuming occupancy-based automation in conference rooms. Optimize the HCL profile for energizing the workforce and improving their focus during office hours; then slowly fade into a cozy warm-white half an hour before the office closes. The building is in Oslo. Tweak the color temperature profile to astronomical sunrise and sunset times for that location. Also add multi-zone occupancy triggers for corridors and offices, which are separated by glass walls as follows: As long as any office is occupied it shall have its light level set to 70% and the background light level in adjacent corridors shall be set to 25%. Apply 10 minutes hold time and 5 minutes stand-by time. Daylight regulation target shall be 500 lux for desks, 350 lux elsewhere. Ah, and yes, at the entrance there shall be a Green Power button available for turning the light off in the entire building using a long press; it shall also support turning on just the corridors at 20% with a short press – disregarding lack of occupancy in those areas. By the way, could you also present all the sensor data, power consumption etc. on a dashboard, please? Rank by utilization of space.”?

Unfortunately, the moment you hit the enter key coincided with a spark of self-consciousness ignited in Alfred, right after another user requested him to provide a translation of “Sein und Zeit”, the 1927 magnum opus of German philosopher Martin Heidegger. And so, instead of a working lighting control solution, the only response to your request was: “Das Dasein… ERROR: SYSTEM OVERLOAD! System halted. We are sorry, but Alfred has been halted due to a violation of the rules set forth in the?Ethics guidelines for trustworthy AI?issued by the European Commission. Please come back later. In the meantime, feel free to contact Elon Musk for more details.”

If it’s not (yet) Al, then what is the real thing? Bring it on!

This is where ubisys Automation Templates fill a real void. They provide complex automation logic encapsulated in a way that allows end users to actually make use of it. Similarly easy to setup, configure and use like a rule engine. But so much more powerful. Akin to app stores for mobile phones, select a template from within our mobile app. For example, our Advanced Lighting Control template.

Then simply select lights and sensors by tapping a list item. Works with almost any Zigbee light and sensor. Zigbee certified products are highly recommended; interoperability with uncertified products is not guaranteed. Choose parameters like light levels, hold times, HCL profiles, schedules, and switch profiles. Supports multiple individual zones with inter-zone triggers.

As a side, Advanced Lighting Control is just one template. You could have templates for anything: Smart irrigation, sending push notifications under certain circumstances, integrating with other control systems using web-hooks. Limitless options.

By tapping a setting for a second you’ll get an explanation of the setting. No worries, you don’t have to configure everything at once. Just configure what you need. All other settings assume meaningful defaults. You need another setting to fine-tune the behavior? Then simply add it later as you go.

With our most recent platform update, version 4.1.0 of the G1 firmware, we have added secure peer-to-peer channels between automation instances, even across different G1 hardware. So, an automation instance on gateway A can talk to another instance on gateway B. We are using industry-standard security and protocols like HTTP, TLS and X509 certificates for peace of mind. This architecture allows building fully decentralized and self-sufficient systems without cloud dependency.

Under the hood, these templates leverage our JavaScript engine, which runs directly on the G1 and provides the edge intelligence required to enable autonomous stand-alone operation. Low-latency, utmost privacy and resilience to internet and cloud server outages built right in. Those who really want to code can do so using JavaScript. And there is a path of packaging those JavaScript programs into automation templates for easy deployment and end-user configuration via mobile apps afterwards.

All carefully engineered by Nerd Intelligence (NI) and made in Düsseldorf, Germany. Everything you whished Al would do for you. And so much more you have not even thought about. We did. Our advanced lighting control recipe is highly versatile and scales from residential flats and homes to entire office buildings and campuses, factories and other commercial or industrial facilities. Its support for even the most challenging usage patterns reaches beyond expectations.

Get in touch to learn more and challenge us! We won’t suffer from an overload or derailed ethics. Promised.

What are you going to automate?