Borbaton

It was a kickstarter post on my Facebook feed where I first saw the clownish product name of Borbaton. So many random products had hit our feeds over the years that I scarcely even noticed it amongst the 80% ads that now propagated my daily view on the platform.

 

It looked like a floating droid ball from a science-fiction movie that might move about and shoot lasers at enemies. AI driven content though had long since made it nearly impossible to tell reality from fantasy. I lingered on the post though curious at the least what the name even meant. 

Bird + Orb + Automaton = Borbaton

The drone-like device promised an impressive array of sensors and capabilities that were meant to aid the owner in finding and identifying more species of birds while in the field. The ad even claimed the product was live and their project only needed start-up capital to produce the first lot of them for field testing by backers. Typical language of, “spots limited”, “the future is now”, and “birding will never be the same again” were plastered across the video sequence showing a user controlling the Borbaton setup via smartphone and voice command.

It was mesmerizing as one user drove up to a wetland and simply held the device out the window and commanded it to scan the entire wetland area for uncommon to rare species. It lifted off from their palm in silence and jetted up to about 100’ feet in the air. Overlaid text and audio indicated the device would create a map of the space to cover and then begin using visual, audio, and thermal scanning to map prospective species. Tools were noted to include extensive machine learning models for visual and audio based identification, which would be reported back to the users phone app. Device based advances were indicated to be nearly silent propulsion and hovering modes allowing near zero disturbance to birds. 

The video shifted to the still seated driver reviewing a phone app that was beginning to show muted listings of common birds in the marsh. The Borbaton hadn’t yet dropped altitude to start scouring the area when a yellow flagged species popped up on the screen indicating a 75% confidence rating in the presence of a Common Gallinule with audio being the initial source for identification. The camera shifted view to the device now descending and silently moving towards the back corner of the marsh space. A few seconds later it sat frozen in air as a chime could be heard from the phone. Panning back to the phone screen showed a video icon next to the flagged record now showing 100% certainty with audio and video proof being noted. 

The user tapped the screen and up popped a live video feed showing a once hidden Common Gallinule slowly paddling along a line of cat tails. The audio feed filled the vehicle cabin with calling Red-winged Blackbirds and the soft grunts of the Gallinule that was much to far away to be heard by ear or without the aid of this particular device. 

The driver spoke again asking the device to continue scanning the area. The advertisement wrapped up with a splash screen indicating 500 spots available for the pilot program and a hefty $8,000 price tag. It would be a $4,000,000 dollar startup with promises of advanced software updates for early adopters to bring on further capabilities in the future. A short list included full autonomous mode, deep scanning, nocturnal analysis, and something referred to as hive mind. 

I pressed the replay option and watched it all again, fascinated by what might be if this were not just a cash grab scam targeted to the birding community. A day later I still hadn’t made up my mind, but returned to the bookmarked web link for the starter that now indicated less than 50 spots remaining with an estimate of only a few hours before it would likely close. I had done some research and found that the project team comprised a couple former Cornell alums and an MIT friend that developed the engineering around the drive mechanism and silent operation features. Beyond the fact it was almost to good to be true, it really did appear to be an honest attempt at bringing autonomous drone technology to bird watching. It would be the next step beyond having listening devices that pull in lists of birds all day and night from a stationary position. 

Little did I realize that upon taking this 8K leap of faith I would enter a world of bird watching I never could have imagined. My own field mobility had waned several years whence, though I still greatly enjoyed getting to the more remote locations. Immersing myself in the sounds of the wild was a joy that helped reduce the isolation and loneliness that permeated most of my days now. My hearing had dulled on one side considerably and back issues finally found me after decades of long hikes. I figured I was in better shape than many my age, but each year a little bit more was being taken away. I longed to somehow stave off time itself for even just a few more years before I might be rendered mostly sedentary. I knew the bell would toll for me one day and all I sought was to continue my long friendship with the birds in any way that I could. In that light the money would either buy some small measure of what I sought or be a lottery ticket dream that simply didn’t win the day. I was fine with either outcome honestly, but I wasn’t prepared for what came next. 

- - -

Weeks stretched into months and despite consistent social media posts I began to lose faith a product would arrive. My assumption was that I had in fact purchased vapor ware and nothing more. In the meantime I had begun reviewing the source code that had been posted to a repo online for the project. It appeared that the developers for some reason decided to open source their project code, perhaps with the intent that it would be improved by the user community during testing. That seemed a stretch though considering how few birders were actually programming experts. The documentation page gave even fewer clues as it could scarcely even be understood that the code was intended to run on the borbaton platform at all. I had found the repository only after a cryptic social media message led me to an unrelated users profile with a truncated URL link. I cloned the code and moved a copy to my cold storage drive assuming it would be a way to recover any future device should the local storage fail. Though I never really thought a device would materialize in all honesty.    

Sometime in the deep of winter word went out that devices would ship in the coming weeks once they cleared approval from the government. During production new regulations went into effect that instituted required testing and licensing protocols for any type of autonomous drone technology that would be fitted with AI pilot routines. Though it was boring reading, the regulations implied that drone technology would not be allowed to run on a software platform fully controlled by artificial intelligence. Everything made me wonder if the end result would be a hamstrung device capable of much, but not all that was promised. I would of course take what I could get if it meant a flight capable device with near silent operation that was still capable of finding birds and identifying them using existing machine learning models for audio, video, and images. 

As the wait continued to stretch into additional months I recognized in the codebase I cloned from the now defunct repository that the pilot core actually appeared to be based upon a modified autonomous vehicle. I quickly concluded that the codebase that had been shared was in fact the original intended code for the platform and that I likely had pulled one of a very limited number of copies. When I performed my pull the website had noted just 6 users total. What all of this meant wasn’t clear beyond the reality that it was seemingly gone along with the user account that had posted the original link to the code repository itself. The developers social media account had been scrubbed of the reference to the now missing user account and in place was now a post indicating a direct link to the open source code was available. 

The two code bases were light years apart from one another and presented what I originally thought we’d see. A hamstrung device platform with limited AI and simple search mechanisms that would likely go off in scripted methods and routes over a defined space. From what I could tell it was still pretty impressive, but nothing compared to the advertisement seen on Facebook. Oddly even that video and ad was now missing from every location I searched. It was as if the origins of the project had been scrubbed completely from the web during the production approval cycle.  

Just like that though a large box arrived via signature delivery in the early spring. It seemed comically large for a device that would fit into the palm of my hand. I had to open it on the front walkway, remove the contents, and break down the box to get it in the door. There it sat though on my coffee table in a black three-point stand held like a crystal ball covered in a fine coated mesh. When I placed it down the weight of it felt substantial for something that was to hover and fly as if defying gravity itself. For the first time in ages I felt those prickles of anticipation and excitement of a new toy a child might feel on the morning of a birthday. The one sheet color guide was really just a QR code to open the real instruction manual on their website that I had already covered a few times over so I ignored it outright. 

I had already installed the core interfacing application on my phone several weeks prior and at this point was ready to go after device pairing was completed. Proximity alone was enough for that and lights began to flicker along with a quick hover check as it gently and nearly silently rose off the stand. After a few rotations the borb settled back to the base, ran another series of checks and popped validation content streams up on my phone for audio, video, and thermal imaging. Device audio asked me to stand up and move about for follow mode. I stood up slowly, feeling my age in the process, and shuffled slowly across the room. As I grasped for a handhold a few times, the Borb moved towards the ceiling and slowly trailed my position while streaming video of the entire sequence of my movement around the open living space of my home.

I wrapped up in home testing with my mind spinning from what I was experiencing even without birds being included. I realized that upon finding a rare bird that a Borb could be sent out to monitor its location as long as it had charge. The ramifications for chase birding were pretty wild with the potential for real time updates to be streamed to other reporting platforms while the finder was comfortable in the car or out bird watching in other spaces beyond the target bird.  I imagined even protected species could literally have a Borb keeping an eye on them and holding aggressive birders or photographers at bay when necessary. I needed to begin field work immediately.

 

- - -

Just a few feet from the parking lot was the first wetland and scrubby wooded space I wanted to try out the Borb. I’d known this space and the park itself for some 35 years now. It typically had nothing of great interest, but as a control I felt like I knew the space pretty well. Trails ringed the edge of the woodland buffer with a few incursion points allowing viewing of the few acres of water, last year's cat tails, and several downed trees along the edges. 

My plan was simple with this test. Send the device out while keeping my own list from my stationary position with binoculars and ear birding only. Let the Borb stream a species list back to my cell phone and compare the two to see how well it does against my own efforts. 

I held the Borb out with my right hand and it lightly lifted off holding position in front me waiting for commands. 

Me: “Scan the wetland in this area stopping at the wooded ring, parking lot, and campsites along the far shore for all bird species possible.” 

Borb: “Command understood, beginning scan mapping.”

The Borb shot up to the expected 100 feet of elevation or so and hovered for about a minute as it presumably scanned the area and generated a map of the territory to be covered. The AI routines I had read were capable of parsing language pretty well and using machine learning models could photograph the area and figure out what I meant by wooded ring, parking lot, and campsites as constraining parameters. 

A chime hit my cell phone and I was shown a map with live images that had been stitched together of the area from the Borb. The map was overlaid with a grid indicating what it had interpreted as the borders of the zone to be scanned along with an additional grid of what it would fly as it transited the area listening and looking for species to identify. I was able to push an agreement icon for the search plan and things began in earnest. 

I then put my phone away and began birding as I normally would, easily ignoring the Borb at this point as I was locked in on my own identification tasks and honestly the device was so quiet and low flying that I couldn’t even see or hear it at that point. It was what I most expected from this location for the time of year. The ice had cleared just a week or two prior and a handful of ducks could be heard, though from my position I couldn’t see most of them, I still knew them to be Mallards. An Eastern Phoebe could be heard calling, and was a early arrival and my first of year for the species. A single Red-winged Blackbird was calling to setup territory as well on the near shore. Soon I heard a Song Sparrow belt out full song as well followed shortly after by a handful of American Crows flying over the pond causing a racket. 

I kept listening and looking for birds and eventually ticket 14 species until the chime came from my phone at the Borb had completed the effort and was returning to base. Seconds later it zoomed up over the dead cat tails and froze in front of me at arms length. I held out my right hand and the Borb settled quickly and shut down leaving all it’s weight in my hand. My phone chimed a second time indicating the command sequence and effort was fully complete. I set the Borb down in the car's front seat after shuffling back to the parking lot and subsequently found a bench to sit on while reviewing the results. I took a deep breath and opened the app for the Borb and selected the most recent engagement and then the option for review complete checklist of species and counts identified during the scanning operation. 

I stared at the list for several moments, stunned at what I was seeing. The list indicated the device as identified 27 bird species and a total of 54 total individuals during the effort. The device had nearly doubled my own efforts and though I expected more I just didn’t think it was really possible from just a single location at this time of year. I was at 14 species and 22 total individuals so I was keen on figuring out where all of the added species were in such a small space.  The list was a expandable for each species fortunately and when I opened Red-winged Blackbird it showed GPS coordinates for 3 total individuals despite my hearing and seeing just one during my own efforts. 

I further clicked on display map for the species and their GPS locations were placed on the original grid map it had generated for the effort. The one bird I had seen was clearly marked in the exact location I had first seen it minutes ago. The other two though were shown off map a few hundred feet away at another pond to the north. Though the Borb had never left the target area, it was still able to map out birds it was able to hear with incredible accuracy using audio triangulation features built into the audio array of the device. I then examined the 13 other species I hadn’t even identified on my listing effort and was surprised to find Tufted Titmouse on the list. I had spent so many years in the park (almost 20) before finding even one short term visitor that I was certain this was a device error. In fact I recalled the problems of the early versions of the Merlin app from years prior where small segments of other species calls would regularly be identified as Tufted Titmouse. As I clicked into the GPS coordinates and mapping they resolved to an even further away woodlot to the north and west. I knew it well also and it was reachable by trail at just about ½ mile.

I girded myself and grabbed a walking stick to help steady my efforts. I was resolved to make the walk even if it took me half the day. I had to see this bird to verify what was happening and what I was being shown. I had an idea though and grabbed the Borb from the front seat of the car first. After holding it out in front of me I asked it to track down the most recent Tufted Titmouse identification and keep an eye on it for me while I walked out the location. The Borb acknowledged and shot up into the sky out of sight. Roughly twenty seconds later my phone pinged with a message that the bird had been found and was being monitored. 

I shook my head, at a loss for what was happening. It was so surreal that I was now following the bird GPS pins of a non-human device. I set out at my own glacial pace that was the standard for me these days. Breathing deep I finally noticed the crisp cool air of a spring morning and savored the exhale. I didn’t know how long I’d be able to get out into these conditions, but they filled me with such joy and appreciation for nature that I hoped it would be several more years at least. Progressing slowly I turned a corn at the end of the wetland and wooded ring of the pond and saw the ¼ mile distant woodlot that promised to hold a Tufted Titmouse. It wasn’t a rare bird for the area, but again it was one that just didn’t spend much time in this particular park and was more likely to hold fast to the river valley several miles east. I checked the Borb app a few times and saw that the tracked bird had moved several hundred feet, but was still within the same wooded space and not far off the natural trail that bisected the space. The Borb app in the meantime had begun a more detailed list of the space it was now monitoring and had jumped up the bird list to now 30 species since it began with the most notable species being a Northern Flicker in the same area as the Tufted Titmouse. I estimated this to be an overwintered bird that was finally free to move about the area since the bulk majority of them had not made it back to the area yet.

I finally entered the woodlot twenty odd minutes later and nearly immediately hear the Tufted Titmouse calling it’s Peter, Peter, Peter call. I half expected a Blue Jay for the number of times I’d been trolled in the past by them masquerading as another species. Not today though as I finally laid eyes on the Titmouse after a few more minutes' walk. Even that was not a normal event though as I neared the prospective location I spoke into the app to have the Borb give me a visual indicator of the bird's location. The Borb moved positions and projected a foot shape on the ground in front of me that glowed and pulsed in a greenish light. Then one after the other like it was walking me to a viewing position. After a few dozen of these, new projections began in red up the side of a tree like an ant march of dotted lines with an arrow in the lead. Eventually the arrow froze and just pulsed and my phone issued the audio note.

Phone: “The Tufted Titmouse is about six feet beyond the final pulsing arrow.”

I picked it up quickly as the bird seemed to peck at a seed or nut it had found while foraging. I was dumbfounded how developers had created such a robust device and how it put me on a bird with such ease nearly a ½ mile away from where it all began. It was in this moment that I realized everything was now different. Birding for the remainder of my days would now be something else that I’d never experienced. I didn’t know if it would all be positive and well changes, but the implications were amazing for what could be accomplished. I had used just two features of the Borb at this point and was beyond impressed with what it could do in those use cases. Just then the app pinged and asked me a question.  

Phone: “Would you like Borb to take a photograph for documentation of the bird?

Me: “Yes, please do.”

Phone: “You can also configure Borb to take photos of all species categorized as uncommon and beyond if you like.”

Me: “Yes that’ll be fine.”

Phone: “Borb configured for automatic photos of all uncommon to rare bird species.”

Just then a photo icon showed on the display next to the entry for the Tufted Titmouse. A high resolution photo of the Titmouse was on screen once I tapped the icon showing better image quality than anything I’d taken in my long history as a bird watcher. I asked the Borb to follow me back to the car and continue scanning for new species on the way. I put in that it should hold position for any rare species, but the return trip was relatively uneventful as we walked and floated together along the trails. The only thing that peaked my curiosity was a ping about half-way back to the car when Borb noted.

Borb: “My sensors have detected a Purple Martin flying at higher altitude above our current position, but you won’t be able to see or hear the bird and it will be out of my own range in roughly. 12 seconds.”

The bird had been noted because it was perhaps a week or two early for arrival. 

Me: “Would you be able to get a photograph if you pursued the bird?”

Borb: “Yes”

Me: “Please do so as quickly as possible.”

The Borb silently jetted off and was out of sight within seconds. I looked at the phone and tapped the video icon in the upper right to show a live video feed. At first I just saw blue, but soon a black dot grew larger until I could see wings and even a light purple hue and a slightly forked tail. The Borb locked in on the bird, adjusted focus and zoom a bit and the image froze showing the photographic evidence of the bird I would never see personally. It would turn out the sighting was in the top five earliest for the area for Purple Martin. I wondered how reporting would be handled considering this was a real record, but not one a human would be capable of producing. I realized though that stationary listening stations had been listing for years so this was scarcely much different. My only moral quandary was whether I should be claiming the sighting as well or not. After all, the device was now an extension of myself and only took the actions it did based on my instructions. Wasn’t it just a new pair of binoculars or a camera with more functionality and I’m the one behind the device?

The day was just the start of the rabbit hole we all went down when the Borb entered the birding community as the first partially autonomous birding companion. When I arrived back at my vehicle I thought of the cold storage code I had cloned from the repo and the fact that it looked an awful lot like it was meant for a fully autonomous device. What if the government had made them remove that code from the devices? Did a version of reality exist where Borb would just be on it’s own and could go bird watching alone, stopping only to recharge? Would I be bird watching with a device that would act more like another person than something I owned? Would I have to ask the Borb to come birding with me?

I would get answers to these questions and more eventually. I hadn’t even scratched the surface yet.