Introduction

Good evening, I'm Riccardo and as Omar said I'm a waste guy. With Nando we are working on the garbage field and to solve this issue that is

basically a lack of knowledge that let me say in this way raise your hand if you know how much paper kilos of paper you produce yesterday some funny one will say zero but for the rest probably nobody knows how much waste which kind of waste we are producing every day and this is a bit of problem and so right

Urban Waste Collection: The Problems to Solve

now I'm going to show you how we are addressing two different type of waste problem from the urban waste collection to the canteen so the food waste application and so let's start with the with the urban waste collection we are

Three Core Challenges: Segregation, Logistics, and Abandoned Waste

focusing and we are working on three main problems the the first one is the quality of the segregation what in Italy we call the raccolta differenziata is segregation that is actually a different process across the the world but basically you have to put the the proper thing in the proper bin and this is

something is not happening the second problem is the logistic so we are talking about fuel we are talking about huge trucks that is moving around the city and also with the two operators that have to they have a huge workload

in let me say in this way they have to go down to the track says that everything works correctly then come up and drive into a traffic jam so it's a

bit complicated and I think that the third one is a pretty Italian thing that is the abandoned waste across the street so we are trying to solve this issue

Using Trucks as “Sentinels”: Cameras, AI Recognition, and Volume Estimation

using the vehicles so the truck themselves as a sentinel so we are installing a camera and the camera actually we called Nando the camera is installed directly on the on the in two different parts actually of the of the truck in this in this position it looks in the in the place in which the waste will be will be I don't know in English collected yes so this is what what the

what the camera see so a picture of amount of waste then we are using our AI to recognize each single item and of course we are we are able to distinguish between the proper object and the wrong object and then the third application is a volume calculation so we are evaluating the volume of the of the waste we are doing this three things to optimize waste management and optimize

waste management means improving segregation quality reducing the kilometers that every day the the trucks are performing around the city and identifying the abandoned waste how we do that and then I'll try to show you a

Turning Monitoring Into Action: Accountability, Engagement, and Enforcement

demonstration so basically as I said we are taking picture but the good thing is that we have a map and in the map we can understand which is the uh owner of the mistake or of the of the proper behavior but however it can be a house it can be a road application so multiple houses

or it can be a commercial um a commercial waste producer but however we are we are linking the data we are getting to those one who are producing the waste and this is the lack of knowledge then we are analyzing we are we are checking the the quality of the segregation and then we can

help the waste management companies or the municipality to do the two standard thing or three that is communication engagement or sanctions so fine and but we can do in a different way because we are saying exactly in which area of the city you have to go to get results and this means also saving time money and

being effective so you can measure the result of what you are doing the the the collection optimization, so the kilometers reduction means that we are measuring the volume.

Route Optimization Through Bin Fill-Level Prediction

1Measuring the volumes means that we are able to understand the filling level of the bin because if we have a certain amount of liters of waste, but we know the dimension of the bin from where this waste comes from, we can understand the filling level of the bin and in this case by doing this kind of analysis day by day we are able to provide an optimization tool that allow our clients to reduce the kilometers they are doing and then

the final application we have an external camera to check if we have something on the ground

and that's it the good thing is that yes nando allows to improve the the waste collection the the waste management 360 degree but it does with knowing the industry and the industry is a very complex industry waste is complicated because there are a lot of waste everywhere so the the good thing is that we are doing this kind of

Why This Approach Scales: Fewer Sensors by Leveraging Existing Fleets

solution we are not the only one that is doing a waste monitoring there are a lot of competitors out there but we are doing with a different approach if we have to monitor a thousand beans our competitor will use one sensor per each bean so it means 1000 sensor we are doing this just using three sensors

because the truck the number of trucks that can be used for serving a thousand and beans are three or four or less so that's the the thing we are doing and

yeah maybe I can just give you a quick idea of what we do before moving to the

Platform Demo Example: Routes, Stops, and Contamination Insights

to the other section so this is a client in Spain I am allowed to show you what we do so this is a place in the north of spain we are serving different places and this is the the the routes that the truck um are doing around the cities so we have we are getting a understanding of kilometers of amount of waste collection and then contamination so if you can

click on a single i don't know on a single stop of the waste truck you can see the ai that is recognizing that things is properly collected or let me see no in this case everything is done correctly but if something is wrong we are analyzing and saying no this is not a good thing and then we have the volume analysis with the volume analysis

we are doing another kind of programs so I'm running a kind of separate demo but is usually is the on the same platform so you can choose your your roots depending on the clients of course so in this case by measuring the volume we are able to predict which bin will be full at that day and so we are advising you

to do the best route around the city to save uh kilometers and fuel and whatever so if this is uh the navigation that the um the navigation platform that the operators see so he knows which kind of bin have to go day by day and that's it so the second application

Canteens: Measuring and Reducing Food Waste

is about canteen the problem is measuring the food waste because there's a lot of food waste in professional canteen there's an issue

Operational Pain Points: Planning Production and Improving Menu Quality

related to the production planning so how many portions of food I have to prepare for tomorrow since I'm not doing receipts I do not have data to understand how many people will be at my canteen tomorrow so this is a very technical thing I'm pretty sure that I was not able to explain but trust me

that the third point is about the quality of the menu so if people like what they are going to eat or not yeah I think that you can imagine that you cannot weigh every day using a scale the ways you are producing into a canteen and we are producing a solution so this is a kind of thing you can see you you

Computer Vision for Demand Data: Counting Portions Without Receipts

You may have seen in IKEA, so there's this kind of thing for automatic checkout. But we are doing this not for automatic checkout, but for getting data.

So for understanding how many portions of each single receipt have been ordered that day in that canteen. so we can have the the data to propose to the to the organization the exact number of kilos of meat vegetables whatever they have to buy to comply with a minimum overproduction rate compared to what they need to do bit complicated

but so we are measuring how many portion you are you are or you are serving that day to provide an estimation of the other day and we are doing this with another tool that I can show you also later if you want and yeah the the food

Quantifying Food Waste and Results

waste is the difference between what you are prepared so how many kilos of things you prepared and you know and what you are serving we are measuring with Nando we are measuring what you are serving so we are we have the information of the food waste and the the goal is to make this this equation equal to zero and

yeah we we had good results here because it's it's a very actually it's a very easy task because it's it's not been measured in basically across the world we are working in Japan in US everywhere there is no measurement tool for this kind of thing so we are getting good results this is one single clients but

Conclusion: Traction, Clients, and Team

we have more and yeah that's it this is the the company we are working more or with 80 different clients across the world as I said from Japan to US and yeah this is also the team we are working with certain number of clients different nation and we raised a bit of money so thank you