Autonomous vehicles are the next big thing in providing transportation solutions that are safer, cleaner and faster. With such vehicles at our disposal, how can we improve the current delivery systems and provide the availability of items from grocery stores on the same day.
The concept of a parallel delivery system, central to which is an autonomous van acting as a carrier for mini delivery bots. The van arrives at the grocery stores and loads the orders into these bots. It then goes into various neighborhoods and releases the bots in certain locations so that the deliveries can take place in parallel.
The prompt for this project was merely a sketch and some vague cues written on a post-it note. We were asked to formulate the right questions to determine the intentions and purpose of the client for pursuing this project. Gradually, through some voice notes and discussions it was revealed that our client was a renowned car manufacturing company, looking forward to venture into autonomous fleet vehicles.
One of the big challenges of this project was to define and constrain the vast problem space concretely for ourselves.
In this four weeks long project for our Interaction Design Practice class, I was a part of a four member team. We collaborated on different phases of the project. I contributed in conducting user interviews and usability testing, as well as in the ideation and brainstorming phase.
1. Constrain and Define
To kick-start the project, we first needed to define the problem space. To do this, we wrote down all the possible types of fleet vehicles. These included:
- Fire Trucks
- Logistics Trucks
- Delivery Trucks
As students, we were tied down by both, time and resources. Within our first two meetings, we quickly narrowed in onto a choice for which we could approach our user groups most conveniently. A few of us in the team had friends and relatives working in the delivery business, and it was easy to schedule our interviews with these people as they were willing to provide their in-depth perspectives. Also, our campus is surrounded by various food delivery businesses, so we could walk in to these places, and gather some important insights for our research.
The next step was to identify a meaningful problem within this industry that we could solve through our autonomous fleet vehicle system.
Primary Research: User Interviews
Our primary research centered around identifying the inadequacies that exist within traditional modes of delivery from both a customer and business perspective. We interviewed delivery drivers, shipping logistics specialists, general managers of supermarkets/restaurants, and customers of delivery services. Because of the variation among the people we interviewed, we received very different perspectives from all of them. This was extremely beneficial in our design process as it allowed us to equally account for these differing perspectives as we honed in on our core problem.
The people we interviewed included personnel from major country wise chains such as Kroger and Target, as well as small local businesses such as Bapu - Bubble Tea and Village Pantry.
We asked them the following questions:
- How does the system of in-store pickups work at your store (for Kroger and Target)?
- Why don’t businesses have a same-day delivery right now?
- Kroger already has an online ordering and in store pickup, but no delivery system because they do not have the infrastructure, such as an app, a fleet of trucks, or a team of drivers.
- Target also has an in-store pickup delivery and distribution center, but it takes more than one day to deliver the items to consumers
- Not all franchises of a business chain within a city allow for in-store pickup service.
- Not all items are suitable for delivery. Small items are suitable for delivery, however, perishable foods are not.
We got in touch with different people, who have had the experience of living in both densely populated urban areas as well as sparsely populated suburban areas and asked them how did they access the grocery stores. We uncovered the following facts during our conversations:
- Grocery stores are not all that conveniently located. In NYC, a person would buy pre-packaged items at corner shops, but would not trust the fresh produce there.
- When parking, people usually find a spot in a pay lot or pull to the side of the road. It’s usually not too big of a deal to partially obstruct the road, since they will be quick. They could not leave their car there, though.
- Customer typically has to drive at least 5 - 10 minutes to get to the nearest store of any kind. Often times, it is even longer of a commute.
- Parking is almost never an issue. Cars are able to find parking spaces on the streets in front of their location or have driveways/small parking lots adjacent to their location.
We also conducted an online survey through Survey Monkey on Facebook and gathered about ten responses.
Can you think of a time when you really needed something real quick, but you could not go to a store to get it because of your busy schedule? Can you please list such items for me (Multiple answers appreciated!)? Ex: Bat, Razor, Tie. etc
Customers usually want to order toiletries (such as shampoo) clothing items (such as socks and scarves), food items (such as ready-to-eat items and fruits/vegetables) and medicines.
We met with and interviewed delivery personnel from different businesses to understand how the logistics of deliveries currently work.
- Delivery personnel face downtime during the job. They would do different activities for leisure such as watch Netflix on their mobile device during this time.
- Many of the people the personnel met during the job seemed as though they had not left the house all day. It felt to them like they were only willing to put up with a minimal amount of interaction.
- For larger apartment buildings and dorms, the customer almost always meets delivery personnel down in the lobby.
- Speed is crucial to the business when delivering. The more deliveries they can push out, the more customers they serve and the more money they make.
- Some people actually enjoy the interaction with their delivery driver. The driver learns their name and sometimes even their pet’s name.
Secondary Research: Consulting Literature
We consulted different resources available online to expand on our knowledge of self-driving cars. Our professor Dr. Hamid R. Ekbia who has a publication on the same topic, Expanding the Design Horizon for Self-Driving Vehicles: Distributing Benefits and Burdens also guided us on considering different aspects and issues that our design might raise in the future. Some of the articles that we consulted were:
We also studied some of the similar autonomous vehicle systems such as ambulances, and for the sake of competitive analysis looked at existing products in the market within the same realm:
It was also important that we showed deference to the ethical issues that these new technologies give rise to:
We combined and deduced the following final insights from the findings of our interviews with the key user groups and secondary research. Throughout our brainstorming sessions, we gauged the efficacy of our ideas on these insights:
Speed is crucial for both the business and customer side of a delivery. A faster delivery means more money for the business and satisfaction for the customer.
Cost and Efficiency
Quick deliveries are presently very costly in terms of the logistics and labor needed for the service. Companies are interested in the idea, but not ready to take the risk.
Businesses would like to expand more, specially into dense urban areas, but the non-avilability of a suitable space is a problem.
Major chains (such as Target and Kroger) would be the primary subscribers of our service as it will supplement their existing in-store pickup service.
People would like to order many different kinds of items to be available for delivery. In general, they don’t mind meeting the driver outside of their apartment/house.
Current delivery businesses use apps to track the locations of their delivery vehicles and to keep their next orders ready in time.
3. Brainstorming and Ideation
We started off by asking ourselves all sorts of wide array of questions. This helped us unleash all the hidden possibilities, without us being held back by the plausibility or unlikeliness of an idea. In sum, they were 25 questions, some of which are listed below:
- What does a delivery driver do?
- Why does a delivery driver only deliver one type of thing?
- How is cargo loaded into a delivery vehicle?
- Does it have to be a car?
- Why not a drone?
- Can the vehicle change shape?
Constraints and Assumptions
We started off with some assumptions and constraints for our initial sketches:
Faster and Cheaper
We started off with the assumption that customers want delivery to be faster and cheaper.
Deliveries are slower because they are dispatched in a serial order rather than in parallel.
Non-Perishable Items Only
Due to the preservation issues of perishable items, we constrained to providing only non-perishable items.
We defined the core, a value that was central to our solution:
Readily Available Goods
Sketch, Sketch, Sketch
Adhering to this very core, we started sketching out our initial concepts. In general, we kept this process quite crude and spontaneous, so nobody would hold back their ideas. All our four team members sat down together and started brainstorming. Each of us drew our concepts on the white board and invited critique from rest of the members. Once others pointed out some major pitfalls in the concept, we moved on to the next person's idea. Sometimes, the new idea would be a spin-off from the older one. After exploring several options, we narrowed down to three concepts.
Going Down The Wrong Path
During our discussions, we identified a new challenge with autnomous delivery systems: The Last Mile Problem. The term is originally derived from the telecommunication industry. It means getting the goods delivered specifically to the homes and businesses from a central hub. This has become a challenging problem for delivery companies such as Amazon, FedEx and UPS, as there is no way of ensuring if the customer is at home at the time of delivery. Moreover, while humans make the deliveries, they perform a lot of complex and nuanced tasks to reach to their desired spot, such as finding an apartment in a complex, opening doors, climbing stairs, greeting, etc. Also, for food delivery companies delivering in dorms and apartments, we thought what if customers didn't have to pick their orders from the lobbies and instead their orders were delivered right at their doors?
We saw a lot of potential for a design solution to be novel that could answer the above questions. Hence, Concept 3 - Personal Delivery Bots became our instant favorite, and we started fleshing out its details as it seemed to be the right solution.
Concept Details - Personal Delivery Bots
Like your own personal vehicles, these bots could be bought and owned by individuals. They could be placed anywhere inside your house on a charging pad, like any electronic appliance such as a refrigerator. They would have a military tank like hull that would help the bots to climb down or up a staircase. We did our research on current robots in the market, and found that companies such as Boston Dynamics have already built robots like SpotMini that can navigate complex spaces within households and walk on difficult terrain outside.
We devised this concept for the future - year 2025. When the current existing personal bots in the market would have improved even more, and would have developed more abilities such as locking and unlocking of doors. These bots were mobile controlled, so the owner could pass an order to the bot from a remote location such as their office to grab some groceries. The bot would be signed up with some predetermined nearby stores to fetch the groceries, where the store personnel would load the bots with the ordered items. Once, the owner gets back home, their groceries would be there before them.
We discussed the idea with our mentors and professors, who made us see that we were assuming too much and becoming too far-fetched with our concept. The project was eventually becoming more engineering centeric rather than design centric. The bot would have to be highly intelligent and would require a lot of complex sensors and algorithms. Also, the cost-benefit ratio was too high, and the price of buying this robot was almost the same as a normal luxury car.
Back to The Drawing Board
With two more days left for the completion of the project, this was a major setback for us. We went back to the drawing board and thoroughly analysed all our research, findings and insights again since any new concept had to be based on the data that we had. This time around, we used the following questions as our lens:
- How can taking out a human from the vehicle be beneficial?
- What ability an autnomous vehicle can give that a human can't?
Luckily, we found our Concept 2 - The Bot Carrier Van useful, as parallel deliveries seemed to be the right answer to our questions.
Our final concept was a fleet of delivery vehicles running around the city that comprised of autonomous vans and mini-bots. The facility is a shared service between various businesses, so that they don't have to buy their own infrastructure or maintain a team of drivers. The service also provides standard sized packaging boxes to the stores so that they can easily fit into the mini-bots. Following are the key entities essential to our system, each contributing unique elements to the overall system:
The system acts as a central node to all the eCommerce systems of the businesses that sign up for this service. This helps in marking their products as "Same Day Delivery" eligible items. It also adds the functionality of letting the users select a suitable delivery time.
It helps in monitoring and scheduling the routes of the autonomous vans by devising most optimal path plans based on distance, time and location It also helps the van find the nearest parking spot for ejecting the mini-bots.
Order Sequencing and Tagging
The system also informs the personnel at the stores in what priority to prepare the orders. It also helps them print and paste codes on each package.
It generates automatic pin-codes for the mini-bots which are assigned to the users through the mobile app at the time of delivery.
At the back of the van is a bar-code scanner. When the personnel loads the packages into the van, the scanner scans the tag to verify that only the right and relevant orders are placed into the van for delivery.
The store personnel just have to place the package into the scanning area of the van. Once, it has been identified that the package is the right one, a robotic hand inside the vehicle picks it up and places it in an appropriate position. The hand also loads the packages into the mini-bots that are residing inside the vehicle.
The mini-bot is securely locked and its contents cannot be accessed by any unauthorized person. It can only be unlocked by the person for whom the delivery is intended. This is done through a pin code that is generated for the customer at the time of delivery.
The bots move on the sidewalks, and their speed is about 3 miles per hour.
The bots have a retractable light/buzzer that the user could invoke through the app in case they had trouble finding the bot in the nearest parking lot. This would help them locate the bot.
Order Same Day Delivery
A same day delivery mark would be added to the existing mobile apps of grocery stores. Only specific products will be eligible for this service.
Users select the same day delivery option at checkout and are only able to buy certain number of items, as the shipping box size is limited.
Select Delivery Time
When same day delivery option is selected, user also has to specify a suitable time slot for delivery.
Central System Gets Updated
Once a request is placed, the central software system notifies the relevant store and schedules a pick up time by signaling the autonomous van.
Order Request is Received
At the grocery store, an order queue is received. The store personnel print the code tags to paste on the packaging boxes.
Order is Prepared
The store personnel fetch the items on the order list and get the packaging box ready for the autonomous van to pick up.
Van Arrives for Pickup
Once the van arrives at the store for pickup, the personnel load the packages where the tags are scanned to make sure only the right orders are carried.
Packages Are Arranged
The robotic hand inside the autonomous van, places the packages in an orderly fashion, and also loads the vacant mini-bots with these boxes.
Van Leaves for Its Destination
The van leaves the store with the package for its destination. The mapping software informs the van of the most optimal parking spot to release its bot.
Asking for Confirmation
Before the van arrives at the premises, it asks the customer if they want their delivery to be made at that specific time.
Mini-Bots Are Released
The system's algorithm decides the optimal location (a parking spot) for the release of multiple bots.
The Bots Take Their Path
Once the bots are released in a neighbourhood, they get onto the closest side walk and start traveling to their respective places of delivery.
A Notification is Sent
A push notification is sent to the customer that their order is waiting outside. At the same time a pin code is given for unlocking the bot.
Order is Received
The customer ventures out to get their package, if they have trouble locating the bot, they can signal it to give them a buzz so that they can see where is it parked.
4. Usability Testing
Our usability tests centered around the use and interaction of the mini-bot in our initial prototype. We set the user up with a scenario and then watched as they completed the end task of removing the items from the delivery drone. We specified tasks of particular interest and ended with some questions regarding the use of the bot and their overall impressions and attitudes towards it as a service.
Users tended to understand the authentication feature of the drone and were able to authenticate themselves in order to remove their items. However, users complained of the mechanism in the prototype used to open the drone to remove items and suggested the hatch would be better served on top of the drone near the authentication portal. Users generally felt the service would be useful and found it appealing, but questioned the cost of such a service.
Our solution was closely tied with our insights. With having a shared delivery system, parallel deliveries and intelligent routing management system we aimed to speed up the current delivery process. Moreover, by sharing of logistical services between business owners we aspire to cut down the price point for such a facility. With such a service running around the city continously, businesses don't have to spend in expanding their chains. We reckoned that our solution would help businesses to improve their reach.
These are the lessons that I walked away with at the end of this project:
Good Ideas Don't Have To Be Novel: We got super excited and carried away with our first idea because we thought it was innovative. This made us loose track of all of our findings and insights. Thus, I learnt the hard way that good ideas don't exactly have to be novel, rather, they solve the given problem in the optimal way.
Good Ideas Aren't Perfect: We spent too much of our time discussing and making one idea perfect in the first iteration. I learnt that time and resources are valuable, and as a designer we have to learn how to constrain ourselves. Ideas eventually are perfected over time and it's a continuous process. It's ok to admit of some loopholes in your initial design proposal.
Keep Multiple Ideas Handy: We as designers should always brainstorm multiple ideas and flesh out the details of a handful of them. This way if we see pitfalls in the solution we decided to pursue, we can always go back, pick an alternate idea and keep moving from there.
How To Design for Experience: Before this project, every design problem that I solved involved designing an interface or something tangible. But through this project, I learnt to design a concept and an experience. It was hard to ask the users about their opinion by making them imagine scenarios.
Team Dynamics: In such a conceptual project, it's important that we keep a check that each team member is thinking in the same direction, otherwise the internal conflicts can result in too many arguments, which eat up lot of precious time.