# [Case-Study] Optimizing Warehouse Asset Monitoring with Docked Drones

- Event: NestGen ’25
- Published: March 7, 2025
- Length: 27:40
- Watch: https://nestgen.org/library/watch/7NRNXEw7jKQ/

> The transcript of a recorded NestGen session, for reading or for use with an AI tool. It comes from the video's
> captions; automatic captions can mishear names and terms.

## Transcript

**[0:21]** well good day everyone and welcome to neston 2025 I'm NAD Berry and my spead business development in the Asia Pacific region here at P I am absolutely thrilled to have all of you here with us and um just want to know how everyone is doing please drop a quick note in the chat box and let me know where you're all joining from and what you're most excited about for today also please do not forget to share your thoughts on our NestGen wall of love using the hashtag NestGen today's session is titled optimizing Warehouse asset monitoring with dog drones as you know this will be covering one of a deployments for tnv aeronine being the world's largest drone solution provider has been you know using Innovative dog drone Solutions and is now revolutionizing the asset monitoring in Warehouse environments they drive efficiency safety and operational excellence aodan group is a global leader in drone technology and data analytics and they deliver Integrated Solutions

**[1:25]** that Empower Industries to optimize their operations the speaker for today is Asar Asar heli here is the technical product manager in the aerodine operations and engineering department he has over 9ine years of experience in remote sensing and a deep specialization in drone technology as Asar is actually shaping the future of aerodine nested drone Solutions his contributions and advancing aerodine initiatives and chid's Mining and energy sectors have been pivotal I also work closely with asper and I know that he is the absolute right person to you know speak on these kind of topics so he will be showing you how drone dog technology is set to transform Warehouse asset monitoring before we begin today's session uh there are just a few house rules that I would want to keep on the table please type any sort of questions that you have into the chat box or the Q&A box we would be addressing them in the question and answer segment that would be post as far presentation this session is also being recorded and will be available for viewing after the event um

**[2:29]** please do not forget to share your favorite moments of NES genen on social media please use the hashtag nesen also a huge shout out to our sponsors um commercial UAV news abss fly freely UAV onic drone land and drone NS and now without further Ado I would want to hand over the stage to Asar who will deeply dive into how dog drones are optimizing Warehouse asset monitoring and driving operational excellence over to you as thank you Nina um so if I'm able to share my screen right now for presentation sure hello everyone um first and foremost I would like to thank you for joining my session today um today we are we will be talking about how do aerodine help one of the leading

**[3:30]** Energy company over here in Malaysia the taga national barad optimize their warehouse assets that is specificate to support um Malaysia electrical distribution and transmission um industry um today like as we as we all know of people who who is very familiar with um modern warehouses would know that a very large Warehouse would have would be very difficult to manage and also it would be very inefficient if things are chose to be done manually right um You will have limited visibility in large warehouses and it's also very time consuming to do all of the individual inventory especially on verification on asset verifications um I will I will address um these challenges today before that let's

**[4:34]** get to know about my company our organization aerodine we are present in about 45 countries globally so far and it is growing um each of the countries each of individual countries or each individual Regional Offices I'm dedicated to several specific Industries per se like in Malaysia we are we are leading in um energy and oil and gas or our Regional Offices in Brazil for example they are doing very well in oil and gas as well um in Chile specifically mining um in USA we are doing good in energy and we provide um drone services like using data drone data and value added with AI powered analytic to solve complex industrial challenges um like we've been in the industri since 2016 and we have a very strong track record of successful successful implementation not only in soui

**[5:38]** ASAS but uh across across the world as well and for this uh for this specific today specific topic we are one of tnb key partners that helping them um doing their striving their technology uh transform transformation so um about tnb tanaga National barad they are serving about 10 million um household of customers in Malaysia because they are the only um energy uh suppliers over here in Peninsula Malaysia um and they operates extensive network of distributions and transmission lines from the north to the south of peninsula Malaysia um and also we also manag this grep and Export some of our some part of our energy to our neighbors as well especially to Singapore and as well as to Thailand um okay and about uh in regards to warehouse management

**[6:44]** tnv is operating one of uh 144 hectares um a warehouse uh in the central region of the country where it serve um the central region of Malaysia like this Warehouse is a 43 hectares and it is splitted into two parts one part of it one on one end um it is handling uh distribution assets and the other end it is handling um transmission assets right so from the previous picture you will see that the area is vast it is uh filled with a lot of uh equipment be IT wires insulators um Transformers switches Etc it is like it is very vast amount of um equipment and U in a lot of quantities so doing um manual inspection will take a lot of time in

**[7:48]** fact when we spoke to our client initially back in 2023 there the initial paino would be the time constraint because for each round of of um asset inspection it will take them about 3 months to complete however the asset movement um for this storage facility it's on daily basis on working day so from Monday to Friday there would be a moving a movement of equipments going in and out so when you are progressing into your manual inspection schedule like by the second week your your initial data doesn't matter much anymore like it'll be it'll not be as accurate um and also due to the large movement of um assets going in Daily um space optimization is crucial because um for equipments like Transformers like um

**[8:52]** a bu of wire they require a lot of space for storage right and um and 43 hectar is not a big of an area to store all of this large equipment hence um it would be very beneficial for the for the clients for our clients to know like how much how many more things can we put um today or tomorrow or the other day so those type of input would really help them manage their warehouse operation uh much more efficiently so our solution um as I'm informed just now um we have been in talk with tnb for this specific project since September of 20123 hence by October of 2023 we have deployed one DJI docs DJI doc one the first version of DJI do that utilizes um DJI matric 30 in tnb Warehouse where we have installed it and powered it with um

**[9:56]** starlink and we we've also uses flight B Enterprise platform and integrate our AWS um um server for flight based Enterprise uploading all those data into our AWS so that we can later download all of these images captured and process it um we've also using um AI that we've developed with tnb like how do we detect each IND individual components like um um how many type of Transformers are there how many types of insulators what are their quantities where are they located um Etc and we we've also developed um a software to visualize um the asset information right so as you can see from this little image just just us

**[10:58]** flying um in the warehouse area and you can see little grid patterns over here so this area we have a segregated or divided it into five sectors where we will fly um these five SE sectors on a daily basis unless um there are range yeah so we have set up five G missions um we initially we would scheduled it um but um across time we notice that weathers um the extreme weather in Malaysia affect affects our mission scheduling sometimes like if we schedule for it to fly at 11:00 a.m. there will be rain or strong wind at 11:00 a.m. so we still go through our operations with with Mission scheduling but we will have our pilots on standby to execute Mission um at hogley yep and daily data captured will be about 2,000

**[12:00]** images for the whole five sectors and all of the captured images um will be stored in a FlytBase and then FlytBase will transfers all those images to our AWS where our data processes will do the the mapping and when automo is generated um we will run through our AI algorithm for post processing and also to visualize all all those informations in our software sorry so we have been doing this consistently on during working days from Monday to Friday since October 2023 initially obviously there will be a lot of hiccups it is a learning process um operation WISE field operation wise is not a lot but building um the AI Al AI detection algorithm that is the hard part because

**[13:01]** because um for some smaller items um it is very hard for the Drone to recognize if you are only doing G missions mapping right so we need to teach the AI to to um to estimates how how much how many equipments that you can see from the image and also estimates will there be any equipments under those equipment like equipment stacking so so that is that is one of the major challenges as well right um and we we were able to achieve about 90% of uh um of detection accuracy so far obviously it is different across the board like specifically if you are talking about Transformers only the the detection accuracy would be 95% but if you're looking for cable bundles the detection would be about 80 to 85% but if you're looking into um insulators it'll be go back at high 90%

**[14:07]** range of accuracy but all in all um we are able to relay um critical information such as um uh clear spaces like empty spaces to tnb or every other day so within like 24 hours we can tell uh to our client that oh you still have about um 20 acres of area left to put something or there's still an empty section over here that you can plan and put something and then um the client's engineer actually took this take this input and plan for their next delivery so that they can plan where the when the stocks moving they already know where do they put the items and yep definitely repeat identification of inventory is part of the part of the uh part of the benefit as well um so that our AI it can easily differentiate

**[15:11]** and um and identify and classify each of the equipment like even so even though uh even so when we are talking about electrical insulators there are a lot of types um and our AI models were able to differentiate it easily and also quantif y it but obviously the quantification parts it is still being improved but right now the accuracy levels is between 80 to 90% so if you can see here from these images this is part of the samples that you can see where the equipment that outline in greens are the detected part or if you can see like some some small some small equipment it's being outlined in gr those are each IND and individual part as well and also one more important thing is that we like I said before we can highlight there are empty spaces where the clients can plan out to put some more

**[16:13]** equipment so um next part so this is part of the samples of uh um of the equipment that we are able to detect um as you can see each of this has different level of detection accur um even cable slap that you can see from here um they they are actually doing some uh uh stacking so there are two equipment on top of each other right so we have to train our AI to to expect this and try to quantify it using DSM data as well to see like if there's any difference in altitude so that it can easily detect whether it is only one unit or two units same goes for cable drum Etc um so future Ro root map for this project um I would say so far tnb

**[17:15]** has been very very uh pleased with the result um We are continuing to improve our detection algorithm and um so that we can detect Much More Much More type of materials uh type of equipment and with much more quantification quality right um and we currently in the plan planning stage of um deploying more nested drones at the same location because instead of doing um have one dock um for instead of having it for warehouse management we are also deploying another dock for safety surveillance for security purposes right we are also planning to deploy to other tnb places for security surveillance as well

**[18:18]** and and soon um right now we are in a testing phase to deploy one dog at a substation substation for a weekly or monthly inspections so so far so this is all my presentations so if I have um any question I'll be I will appreciate if anyone have some question or maybe n if you great no definitely thank you so much um Asar for all the insights that you were able to provide us you know um we did roll out a few polls during uh in the middle of the session and people joining in were tuning in and you know voting in the polls we are now open the floor is now open to questions and answers so everyone who's joined the session please feel free to ask questions in the Q&A

**[19:18]** section um aswar will be able to answer them for you oh you ran a poll about what is the biggest technical challenge in deploying do drones for warehouse asset monitoring I would say right now our biggest challenges would be indoor indoor warehouses like to me if we can crack the codes and find one perfect solution for IND indoor Warehouse um management that'll be very good because currently we have a limitation with the drones itself isn't it where it requires uh GPS uh GPS positioning so that you can U properly met out map out and properly navigate in a uh in a space where we call it is a gpsd night environment which is very hard to fly in I hope someday that we can see a technology where the Drone uses a slam slam and ligher and complement it with a very high resolution

**[20:24]** camera with an ample lighting so that we can help our client to to help them manage their warehouses as well all right uh thank you so much as well we've got question answers now coming in in the Q&A section the first one is what primary regulatory challenges associated with deploying dog drone systems and how can these be addressed the primary regulatory challenges um I'm not sure I'm not quite sure about any other countries but um in Malaysia specifically we have a very restrictive regulation about flying drone in a manner where you require to operate the drones Beyond visual line of sights so so um a lot of the civil aviation um Authority in the world will have a pretty much a very restrictive um measure on this type of operation um unluckily for

**[21:27]** us in Malaysia it's much more strict where we have to go through like a 5 month subjective um audit process so we need to demonstrate to civil aviation that um that everything is safe to be conducted um especially the flights um the flights will not disturb um current operations of people in the v vicinity like it will not impact on the health of the local people it will not suddenly drop on someone else so and also that um pilot will have will always have awareness right so I would say I would recommend that you need to check with your local local Authority for them to help you with but um if when you crack the code you got your procedes to deploy this um it will be like a very rewarding technology to be used and on a daily

**[22:28]** basis right thank you so much for that answer we've got uh the next question that says how is Aid driven predictive maintenance enhancing the reliability and efficiency of drone enabled asset monitoring how AI predictive maintenance enhancing the relability and efficiency of drone enabled um access monitoring um okay I would say with a dock uh uh a do drone or a nested drone some other names these drones because you are not relying on on an actual operator on site so the operator can sit somewhere comfortable in an office or you can do a scheduled Mission and that drone actually can do a lot of work in a single day like it has a very high productivity but when you your drone have a high productivity usually you will have a bottleneck in your processing part of it like

**[23:30]** you can take 10,000 image in a day but then there's no if there's no software um that can help you process that 10,000 image um the the data captured by the drones will not be usual it has no value value addition to the to the to the pictures so I would say AI um AI driven predictive maintenance or whatever algorithm that you want to include in your processing pipeline um with AI it can help you generate much more useful outputs but in in our case in our case or in this tnb case um it has helped us has helped tnb to to improve their asset verification by tenfolds honestly like without it like without it a normal person will have to go there and do individual verification or even if you want to rely on on a mapping on a daily

**[24:34]** mapping and then let uh G analyst do the identification on a daily basis I would say one single map will take you at least two weeks for you to to to do the whole um classification but with AI we can take the the pictures today and then tnb will have the result by the end of the next working day and by the third day they can use that insights already wow I think thank you so much as well you know that's pretty good to hear we've got one question uh I think the last question out of the Q&A that have been asked what strategies can organizations use to integrate these Cutting Edge dog drone Technologies with traditional asset monitoring H that's a good question I would say um what strategies to me it always you do not plan out your deployment in like let's say you have like 100 hectares of

**[25:36]** area and then and you directly buy like 50 docking station and start operating it um that will be for us it will be a very wrong strategies my recommendation to you is start with one unit first deploy it for 3 to six month and see how how has it added value into your operations and then and then look at add the value and try and you can try to scale it up maybe go for three docking add another three docking stations and and see how it scals up again maybe by the end of the year you will probably have five working do docum station at your size maybe even 10 so use cases um use case of the docking station is a very important over here and and you need to think from a perspective that uh that each individual doc um is one worker honestly that needs a lot of human hands to support for it to

**[26:43]** operates no definitely I completely agree with the last thing that you said last statement that you made especially um thank you so much aser for answering these questions and um also thanks a lot for you know coming and sharing this with all all the attendees here in the session um very very well put across um thank you to the attendees for asking these questions so on and so forth and for being present uh in this session I yeah I just hope that everyone has a good day and great week ahead thank you so much okay thank you n goodbye thank you byebye for