I guess it depends on what you mean by sharing a platform with a consumer version. There will definitely be a lot of commonality in the architecture annd technology, and a modified Gen3 consumer car could certainly make a decent robotaxi. The modular unboxed manufacturing process and structural battery pack help make it easier to add variety to the parts of the vehicle above the skateboard. Still, a robotaxi fleet comprising minimally modified consumer cars would be far from optimal when compared to a family of dedicated robotaxis designs.
Specialized solutions, in their niche, will always outperform generic solutions. The more you can narrow the design requirements and objectives for a system, the more you can optimize its design with customization. A robotaxi has many fundamentally different marketing considerations than a consumer vehicle has. Some key differences include:
- 3-10x more usage per day
- They will either be "bought" internally by Tesla or bought by third-party businesses, but not by retail customers
- Exterior styling and sporty performance less important; cabin experience more important
- Most will never leave their local service area
- Instead of an individual serving all the various needs of an individual person or household, a fleet of vehicles will serve the collective needs of a whole local market
- Most of the time, occupant ingress & egress will occur from the curb side (right-hand side in most markets). Also, disability accommodations are necessary.
- Many of the safety regulations for human-driven vehicles do not apply
- Service will be performed by fleet management teams, not individuals and third parties
- Speculative opportunities:
- Cargo variant
- Higher speed limit
- Optimizing for operation in Boring Company Loops
These high-level requirement differences will translate into major differences in design.
The vastly greater usage rate profoundly affects the economic value calculus. It is easier to justify greater upfront capital expenses for a robotaxi than for consumer cars, because the payoff comes much quicker. For example, suppose the engineers are choosing between two designs. One design is less expensive to produce, but it only has 250k miles of useful life. The alternative design would cost $10k more, but it will reduce lifetime per-mile maintenance costs and will have a useful life of 500k miles. Most retail car buyers would prefer the first design if the savings were passed on via a $10k discount, but the second design is better for a robotaxi that will be used for 100k miles per year.
Consumer psychology factors into this as well. Traditional car design is largely based upon appealing to (irrational) consumer preferences. For example, this is the only reason why so many cars today have enormous front grills, which exist only for appearance, not functionality. Retail buyers focus on factors like style, perceived safety, performance, driving feel, and perhaps most importantly for many customers, what the vehicle will communicate about their identity and social status. People also tend to strongly discount the future and focus more on short-term costs and benefits. Businesses operate more on long-term planning and spreadsheet models. They will look at depreciation, amortization and maintenance much more rationally than a typical retail customer. Durability is just one example of this utilization-rate effect on the design tradeoffs. I expect people will have different psychological relationship with a robotaxi they're riding in, such that cabin experience, price, and service convenience become the more dominant considerations.
When conforming to conventional external aesthetic norms no longer matters much, what else changes? Let's start with a throwback to a first-principles design from GM, the old EV1. This is likely closer to what a two-seater robotaxi should look like. That weird-looking fairing over the rear wheel well improves aerodynamics, and so does the bubbly body shape and the taper on the aft end of the car. This design is excellent for aerodynamics, which translates to better energy efficiency and less cabin noise, but most retail customers would say this looks ugly and dorky.
View attachment 1036451 View attachment 1036454
If Tesla expects that robotaxis, due to their safety and millisecond-scale reaction time, will eventually be approved for top speeds much higher than we legally allow humans to drive on public roads, then aerodynamic efficiency becomes even more important. Additionally, if Tesla expects the Boring Co concept to be successful, it could be worthwhile to tailor the design to serve that use case better.
While styling and exterior aesthetics will matter less, the cabin experience becomes much more important, and the design objectives for the cabin differ too.
- It needs to have better night lighting
- The suspension and tires should be tuned more for comfort, NVH damping, and stability, and less for sporty handling performance and driving feel
- Extra insulation material could be added to further attenuate NVH
- It also could make more sense to give the robotaxi a fancy, Model S-like audio system
- The back seats will be more important because they'll be used far more often than in a consumer car
- Productivity features like tables and electrical outlets might be worth including
- The cabin needs to be designed for flexibility to accommodate a variety of seating configurations (and maybe cargo configurations too) without compromising overall cost and complexity too much
For a $25k consumer car, this stuff simply does not fit in the budget nor in the design priorities, but, to reiterate, for a durable, high-usage robotaxi there is more budget for cabin upgrades without increasing average cost per mile by much. Teslas, and most EVs, have thus far been marketed with performance and driving excitement as one of the main selling points, whereas robotaxis will be marketed more based on the quality of the cabin experience.
Because most robotaxis will stay local, most of them will not need a large battery. Most EVs today have far more battery capacity than is used on a daily basis. The capacity is sized mainly to allow for occasional long trips and for convenience of customers who aren't able to charge at home or at work. For robotaxi fleets, this is a major optimization opportunity. The design of the robotaxi platform will likely include variants with batteries much too small to be competitive as consumer EVs. Transportation demand fluctuates throughout the day, week and year, but especially daily. Fleet capacity will need to accommodate the highest-demand periods. The rest of the time, a substantial portion of vehicles will be unnecessary and will sit idle, serving only the morning and evening rush hours. Therefore, cars with 50-100 miles of range may be useful for this niche. Especially now that Tesla makes structural battery packs, this could affect chassis undercarriage design.
The battery size variation can only work because the fleet of vehicles will serve the collective needs of a whole local market. This new design optionality will also affect vehicle sizing, seating arrangements, luggage space, and even whether to design for a cargo variant to serve last-mile and low-demand delivery routes. Balancing all the needs of a single customer with a single vehicle severely constrains the design options and requires major compromises. The biggest compromise is seating capacity. Almost all trips are with one or two occupants, for an average utilization factor that's probably around 20-25%. Robotaxi networks should be able to perform about 2-4x better on this metric, if the family of vehicles in the fleet is designed specifically for this purpose. As
@MC3OZ noted, the two-seater also could be designed for a narrower body to further improve aero drag, weight, size and cost. If narrow enough, two-seaters could even hypothetically drive two across in Boring Co Loops, which are just barely wide enough to make this a plausible possibility.
Most of the time, passenger ingress & egress will occur from the curb side. Due to this asymmetry in typical loading/unloading patterns, it might make sense to have a laterally asymmetrical vehicle design and unconventional door configurations. Should there be bigger apertures on the primary entry/exit side? Should there be a sliding door like a van or dual sliding doors like the Zoox robotaxi prototype (see image below)? Rear doors as suicide doors (i.e. hinge on C-pillar instead of B-pillar) in order to enable a large single opening?
View attachment 1036468
Also, at least some of the robotaxis will need to provide accessibility for disabled people. The design decisions for the doors and surrounding structure and wiring also may affect design decisions regarding wheelchair ramp integration.
There are even more questions when we take it to first principles and look at the details. Should the low-speed, high-capacity, urban vehicle be longitudinally symmetrical like the Zoox so that it can drive in either direction equally well, to increase flexibility in tight spaces and save precious time on turning around? If we design for aero efficiency without side mirrors and with a radically unconventional rear end design, how does that affect the optimal upstream aero design of the front end? Since the driving is done with only cameras, do we still need a wiper for the entire windshield? If not, how does that affect the rest of the vehicle design, from the slope of the windshield to the arrangement of internal subsystems? How does deletion of the steering wheel and wiring (looking at Cybertruck drive-by-wire as the current state of the art) affect the configuration and layout of other subsystems? When you aren't constrained by driver visibility angles, how does that affect the design of the A pillar? And so on.
None of this stuff should be considered in isolation, because there are all sorts of interdependencies in the design. It's not just about deleting unnecessary parts and processes. It's also about exploiting the new opportunities that arise when those parts are no longer taking up space and constraining other design choices. Good vehicle design requires intelligent systems integration, which Tesla has always focused on and excelled at. I expect they will continue to do so with the robotaxi platform, and that's why it probably will have major differences from consumer vehicles.
