Automotive - Industry structure, form of the car of future et al.
Next: Schrödinger’s Automotive Company
TL;DR Self-driving cars will collect enormous amounts of information about their surroundings and they will travel all roads everyday creating a near real time digital twin of the world. This creates enormous opportunities and risks. They will change the interior of the cars, eliminate many jobs and transfer others, change city structures and to lesser degree buildings and apartments.
After taking a high-level sweep at construction, let’s take a similar, short look at automotive industry. How does self-driving change what cars, houses and cities of the future look like (or could look like)?
Industry Structure
Now car companies are investing a lot into self-driving technology and electric drive. This is due to new entrant from outside of the field. Before that for decades the industry was moving at glacial speed.
In chapter on construction, we discussed how the relationships between industry partners often define how flexibly companies can operate and innovation. Same power is present here.
Tradition car manufacturers are assemblers – assemble-to-order is their main business process. All parts come from suppliers who design and manufacture components such as anti-locking brake systems, seat belts systems, seats etc. To drive down costs and stay in control, the car companies use divide and conquer technique. For example, the ESP (anti-lock control and traction control) in any given car model is given to two different suppliers and the customer does not know which EPS is in their particular car.
A navigator can be divided so that map data comes from one source, navigation software from another and the hardware from third party. This should in principle not matter as the brand is responsible for the whole, but it dictates how the industry can move. Every component in a specific model comes from two source and manufacturers have a sizeable portfolio of different models. The end result is combinatorially rich. When a manufacturer needs to introduce some large change that is cross cutting like changing from combustion engine to electric drive, it means changes all around. Many interworking components need redesigning. Since each model has more or less independently chosen suppliers and supplier base is divided, this means that a whole fleet of vendors need to do the needed change at the same time in order for the new model to reach markets in time.
To add insult to injury, companies may have made very long-term commitments to supplier for cost reduction in some cases - for example to purchase a specific transmission oil from a partner company on a 20-year contract.
Another issue is the operating model. Let's assume that each model has a well defined leader – a product manager who decides what goes in. They are well educated, energetic and want to make a good career. Every now and then there is opportunity to make a big shift in technology and gain long term advantage. But they know that new tech development is always time consuming and risky, time tables are broken and costs explode. This is the nature of innovation. End result is that if the product manager wants to build the future for the company and focuses on creating differentiating technology, their project will be plagued with problems causing extra costs and delays. In simple terms, they will ruin their personal career. A smart person lets someone else take the heat.
Big changes require commitment from the highest level of company and cannot follow normal product development methods. Very few companies can do this as they are aligned to produce good results every quarter. Their shareholders expect this and management that fails to delivery good results gets quickly replaced by a more sensible people who focus on short term results ultimately driving the company to its destruction in longer term. This is what the shareholders want.
These are some of the reasons why large companies with almost endless resources regularly struggle to make even smallest baby steps.
Self-Driving Car Technology
Nothing else but existential threat works and it can only come from a new entrant. The car industries are currently on that journey for electric drive and self-driving. This requires a bunch of new technologies and changes the nature of cars from mechanical products to largely software products.
What type of changes that means?
Cars will host the sensors and logic onboard to enabled self-driving. Networks and services on the cloud can have assistive role. This allows the self-driving functionality advance fast as there is no dependency to any external source. The role of network services is to augmenting fully independent car functionality. Such network services are among others updates to map data with real time feed of driving conditions, entertainment, software upgrades, predictive care (knowing before that some component needs repair), data connection for working while commuting and using vehicle as data gathering platform.
To enable self-driving, the car needs a suite of sensors – such as cameras, LIDAR (laser-based radar), GPS, inertia measurement unit (which uses gyroscopes and accelerometers) and normal radar (to see in fog).
Sensor fusion (combination of data from the multiple sensors) and embedded logic allow the car to locate itself, plan routes, recognise objects near it and to track and predict what they are going to do. Based on that the self-driving logic makes decisions.
Cars will benefit greatly from a high-speed network connection for a few reasons. Firstly, many people will be afraid to step into a shared self-driving vehicle because of security worries. Cameras provide a layer of security for passengers, allow rapid action if needed and ease pinpointing any culprits.
Second practical reason at least well into the era of self-driving, is that there will be corner cases when the car cannot figure out what to do. Special road or weather conditions, failed sensors etc. When the car does not know what to do, there needs to be remote control possibility so that someone can drive the car back from distance to area where it can either be fixed or it picks up its bearings and can motivate itself to drive again.
Third is related to safety. With high speed connection, cars can communicate hazards to each other – either directly from vehicle to vehicle or via a low latency network. Such conditions are for example wrongly parked cars in the middle of road, fallen trees, black ice on road, local heavy rain or hale etc.
Out-Facing Sensor Platform
End-results is that cars will be mobile sensor platforms roaming around and this will be a big opportunity for car makers. If you think of the global fleet of vehicles, they drive every day all roads bypassing every location where humans live. Adding sensors to vehicles allows to collect a real-time data feed of among others:
Road conditions such as icy roads, water on road, pot holes, detect road works, new roads, roundabouts, bridges, tunnels being constructed or opened, state of road markings etc.
Record travel times to build predictive models for traffic. Can also be used by city planners in simulations what different alternatives and ways of organising movements in city could mean. Essential in planning infrastructure investments or in reacting to disruptions (think large water leakage in a metro station).
Inventory all traffic signs, traffic lights and how they work, ads alongside roads
Weather information such as temperature, humidity, brightness of sky and amount of solar radiation, snow, rain, hail storms, lightning etc.
Air quality data like different pollutants in air and physical particles
Every building alongside roads, new constructions of buildings, new industrial facilities, read signs about upcoming construction, who is building and when it is expected to be completed. Part of building data is the style of it. Cars can inventory different architectural styles used in the city and this can be combined with other data such as resident health (if they opt-in) or price of houses to see how the build environment affects us. A quantitative cost can be attached to different architectural styles.
Every commercial activity along the roads (cafes, shopping malls etc.) and how they change and how their business changes. This in turn allows commercial analysis. For example, following how many cars park outside of a supermarket stores allows to predict sales and hence how their stock value will develop. Same info can help new entrepreneurs decide where they should set up their shop, café or restaurant – where competition is lowest with the right customers moving about.
All makes and models of cars in traffic telling popularity of different brands and how competitors are doing.
All accident, burglaries, police and paramedic activities in city, number and quality of green spaces affecting property value. Car companies will have the capability to predict how apartment prices change.
Amount of outside walking, biking and active in different outdoor sports like playing basketball, tennis etc. outside. Tells about physical activity in the city and how the city views general health issues.
By looking deeper into the pictures captures it becomes possible to analyse what fashion brands are gaining in popularity in each demographic group and which are losing ground. What styles in fashion are gaining popularity.
And many more no doubt
Cars will scan all area inhabited with people every day, in most places all the time allowing real-time analysis, a possibility that can be both beneficial and scary if used wrongly.
This data will first be used to help vehicles. Cars that have just passed by the patch of road can communicate what’s ahead to my car. Vehicles will gain collective vision – being able to see past the corner and know what’s hidden behind the hill.
They can create a complete digital twin of the outdoor world that can be sold on public data market places or the car companies can set up subsidiaries to analyse and sell results. Who owns this data will be an interesting question? The owner of the car, the car company, the data collector, do cities want access? And at what point should anonymisation of people be done along the data processing pipeline?
Inner-Facing Sensor Platform
For passenger vehicles inner facing sensor platforms are today used to monitor driver fatigue, noise etc. In future the use will change as people will no longer need to drive. But as sensor prices keep going down in price, its likely that the number of sensors will rather increase.
The inner sensor platforms can be used for tracking temperature, humidity, thumps and shakes, indoor air for certain chemicals and allergens. Allergic people will not need to use ride-sharing robotaxis where the current or previous passengers or their pets have left unwanted substances in air.
Different type of use cases can be supported. Say someone wants to sleep during their morning commute on a self-driver, the route planner and inner sensor platform could be used to plan optimal route or if they have back problems, to plan an optimal route to avoid any shakes + adjusting suspension to minimise it.
Form
The car can be completely re-through when cars drive themselves. Today they are for many users a signal of wealth or personal preferences.
When a car can be summoned in a few minutes and it takes you anywhere, the need to own one diminishes. Comfort and functions inside the car will drive the designs rather than external appearance or engine sounds or the smell of a new car. So, chrome fittings, low profile tires, dramatic contour lose their meaning. Cars become like London taxis – easy to get in and comfortable to sit in with space for activities. People will spend their time inside car working, gaming, on social media, on dating sites, perhaps eating and drinking or talking with other passengers if such a quint habit survives the age of smart phones.
Cars becomes horizontal lifts taking people from door to door. Few want to own their personal lift or pay thousands extra for shiny aluminum railings. As always there are exceptions - shopping malls, hotels and other public spaces want the interior of their lifts match their brand. Likewise, when robotaxis are set up by car companies, they may well want to create unique experience to boost their brand.
Experimentation with the interior becomes possible. Is there a need to see outside of the car? If not, the whole inside of the car can be fitted with displays making the car into a huge mobile VR goggle for games and entertainment. Take roller coast drive inside the car when going with kids to an amusement park or swim inside a coral reef if going to an aquarium. Or study in such an immersive environment like human anatomy moving inside a virtual body.
Augmented reality glasses are a competitor to this vision. In that scenario car interior will be developed to be as neutral as possible to serve the glasses.
Will the car have breakfast facilities and how is sleeping organised? The big question keeping designer wake at night is whether a loo or a shower is a necessary feature for the car of the future. Hair dryer goes without saying.
Assuming accidents are rare or non-existent with autonomous cars, the body of the car can alternatively be made lighter and cheaper. Cars might become transparent so visibility to and from outside is unrestricted.
Vehicle can be split to the self-driving platform and payload part. When car is not moving people about, it can perform other logistics activities like shipping packages. A personal car is however not very good at packing a lot of packages in. The car could be split so that the cabin carrying person is replaced with logistics cabin during off hours.
The self-driving frame could be owned by one company and the payload modules owned by different set of service providers. This requires that the different parts have compatible interfaces. This is however unlikely as a modular ecosystem approach is more complex and thus expensive, limits the control of participants and has even potential security risks.
The splitting would however allow cabin makers to evolve their designs independently and to target smaller audiences - typically increasing speed of innovation. Different ones for entertainment, work, rest, meditation, medical etc. A typical car design process takes five years to seven years, cabins could evolve much faster. Universities could design their own cabins full of sensors for research needs rather than use duct tape to attach sensors to their fixed-form vehicles.
As another direction new forms like an electric bike with small cabin can evolve. Small mobile pods have less mass to move around, cheaper to manufacture and travel at lower speeds. All this result to much lower price point for customers. In terms of function, they can have everything that a car offers on short distances at a lower cost.
In the next post we discuss how a new entrant can use ecosystem approach as a way to appear much bigger than they are and other more manufacturing related topics.