At rideOS, we provide a universal mobility-as-a-service (MaaS) platform that has all of the underlying technology our partners need to run their own mobility network. As rideOS continues scaling with more partners, one common question that our prospective partners have is: “how will rideOS’s solution improve my business?” We are confident that our algorithms are significantly better than other products available in the market today and we have done simulation work to demonstrate our results. In particular, we have found that in addition to showing potential partners what we have done, it is beneficial to show them how rideOS can improve their services.
tl;dr — We are confident that our rideOS UniversalMaaS platform can suit your needs, and we are happy to run a simulation with you to definitively demonstrate our benefits.
We have created the Osiris simulator, and in this article we describe some of its features. We will then share some of our simulation results to illustrate how the Osiris simulator can help you decide to use the rideOS MAAS platform.
What does Osiris simulate?
Since simulation is used in a wide variety of domains, let us first define what kind of simulations are run in Osiris. The Osiris simulator runs a microsimulation of vehicles. Below is a diagram describing the flow of an Osiris simulation.
Figure 1: Osiris simulation
Every simulated vehicle has its own pose, plan, and route. The vehicles move along the road network following real-time traffic, as they follow their routes and complete their plans. As time passes in the simulation, new tasks are requested, which get assigned to vehicles, and the vehicles update their plans and routes. In particular, while the input Osiris scenario contains the information of when and where new tasks are requested, this information is hidden from the rideOS platform services, so that the simulation is “fair” and no future knowledge is provided to the optimization algorithms on our platform. At the end of the simulation, metrics are computed and returned.
How is Osiris used?
The main purpose of the Osiris simulator is to show our partners how the rideOS platform can improve the performance of their fleet of vehicles. To fulfill that purpose, Osiris is able to:
- Run simulations on the rideOS platform using partner-provided / publicly-available data;
- Provide metrics of the simulations;
- Compare metrics of the rideOS platform against benchmarks (e.g., partners’ algorithms, greedy algorithms);
- Visually playback a simulation.
In Figure 1 above, the input to Osiris is a scenario. We have a standardized format for our partners to provide real data of their vehicles and tasks, which we then use as input to the simulation. If partners do not have existing data for us to run simulations with, they can also specify parameters of the simulation for us to run with, e.g., the number of vehicles and tasks in a geofence. In addition, we can also create scenarios using publicly-available datasets, e.g., the New York City TLC dataset.
After the simulation has completed, we compute metrics, such as task wait and active times, utilization, etc. These metrics are then saved and shared with our partner. We can also compute the same metrics using the partner’s dataset, which then allows us to directly compare the performance of the rideOS platform against the partner’s existing algorithms. We will share some of our simulation results in the next section.
In addition to the metrics, the Osiris simulator also saves a log of the simulation. Among other things, the simulation log allows us to create a visual playback of a simulation, which our partners have found very useful. Figure 2 below is an Osiris simulation using the NYC dataset.
Figure 2: Osiris simulation using NYC taxi dataset
Osiris can run simulations of ridehail, delivery, and hybrids of the two — as rideOS provides a MaaS solution for people and things, Osiris is able to simulate people and things as well!
What does Osiris output?
When an Osiris simulation is run, we compute metrics from the simulation, including:
- Utilization — a vehicle is utilized if there is at least one passenger/item on board
- Trip wait time — time elapsed between requesting a trip and a vehicle arriving for pickup
- Trip active time — time elapsed between pickup and dropoff
For example, in the NYC taxi simulation (Figure 2), we computed the metrics to be:
- Utilization: 47 ± 27 %
- Trip wait time: 6.8 ± 4.1 min
- Trip active time: 14.4 ± 12.1 min
In addition to computing the metrics from an Osiris simulation, we compute the metrics from the partner’s dataset as well, which allows us to do an apples-to-apples comparison of our simulation vs the dataset.
Below is a chart showing some of the improvements of the rideOS platform compared to the one of our partners.
Figure 3: Utilization improvement using rideOS platform
The partner provided us with a dataset of trips from 10 of their self-driving vehicles. These self-driving vehicles had map and routing constraints that we represented with our Constraint Data API, allowing us to optimize with their constraints, and also simulate their vehicles accurately.
Through the Osiris simulation, we demonstrated that the trip wait times and active times were similar, meaning that the overall rider experience was not affected. In particular, we were able to increase their utilization from 29% to 48%, using only half of their vehicles! Thus, by using the rideOS platform, our partner is able to half the number of active vehicles, or equivalently, service twice as many passengers with their existing number of vehicles!
We have run many simulations for our partners, and provided them with valuable metrics. Overall, we have found that we can significantly improve our partners’ utilization and trip metrics by using the rideOS platform.
How can Osiris help you?
If you are interested in the rideOS platform, feel free to reach out to us!
If you are interested in improving your existing ridehail/delivery service, we would be happy to run an Osiris simulation for you, and demonstrate how the rideOS platform will improve your services!
If you are interested in starting a new ridehail/delivery service, we would be happy to run an Osiris simulation for you in the geo-area you’re interested to operate in. We can also help to find the optimal size of a fleet for you!
We are confident that our rideOS UniversalMaaS platform can suit your needs, and we are happy to run a simulation with you to definitively demonstrate our benefits. Contact us today