Automobile manufacturers, suppliers, dealers, and service providers understand that quantum computing will eventually have a major impact on most every aspect of the industry. Daimler, Honda, Hyundai, Ford, BMW, Volkswagen, and Toyota all have some form of quantum evaluation program in place.
Classic computers cannot solve many of the significant real-world problems because of computational complexity or because calculations would take an inordinate amount of time, perhaps hundreds, thousands, or even millions of years. Quantum computing offers the potential to solve these problems in a reasonable amount of time. Although current hardware isn’t advanced enough to support the number of qubits needed, we are already working to implement error correction solutions in order to build fault-tolerant quantum machines.
The same hardware and error correction constraints limit the full potential of quantum machine learning. In some instances, it has proven to be helpful with current quantum computers; it can also exceed the results of some classical models.
IonQ has a research history with quantum machine learning, so I was looking forward to talking to Peter Chapman, CEO of IonQ, about his partnership with Hyundai Motors.
First, Chapman explained that the partnership’s goal is to determine quantum computing’s potential to provide improved mobility solutions for autonomous vehicles. For these projects, IonQ will use Aria, its latest trapped-ion quantum computer.
IonQ combined its quantum computing expertise with Hyundai’s lithium battery knowledge two months ago. It is developing sophisticated quantum chemistry simulations to study battery charge and discharge cycles, capacity, durability, and safety.
As an evolution of their relationship, the IonQ and Hyundai team will develop quantum machine learning (QML) models to detect and recognize traffic signs and identify 3D objects such as pedestrians and cyclists.
Recognizing traffic signs and identifying 3D objects are critical elements of Advanced Driver-Assistance Systems (ADAS) used by autonomous vehicles. ADAS depends upon cameras, lidar, radar, and other sensors for inputs to onboard AV computers that interpret and respond to the driving environment. A 2016 study by the National Highway Transportation Safety Administration found that 94% to 96% of accidents are caused by human error. With quantum enhanced inputs for ADAS, it is likely that human error can be minimized to reduce accidents.
Early in his career, Chapman served as president of a Ray Kurzweil company, where he gained machine learning experience. As a result, he has a deep knowledge of classical machine learning models and the complicated steps needed to identify images. More importantly, he understands why QML will be much faster and more efficient than its classical counterpart.
“QML doesn’t need numerous processing steps for traffic road sign recognition like classical approaches to object detection,” he said. “Quantum recognizes a sign and interprets its meaning in one single step.”
IonQ has already completed the difficult computational part of the road sign recognition project. It has already trained quantum machine learning models (QML) using a standardized 50,000 image database to recognize 43 different classifications of road signs. Next, IonQ will test its QML model under real-world driving conditions using Hyundai’s test environment.
Chapman also explained why he believes quantum machine learning and object recognition will prove much more powerful than classical.
“What happens if your car sees something that it has never been trained on before? Let’s take an outlier case, such as a person with a triple-wide stroller, walking two dogs on a leash, talking on their iPhone, and carrying a bag of groceries. If the training data had never seen this scenario, how would the car respond? I think quantum machine learning will fill in those gaps and provide a known response for things it hasn’t seen before.”
IonQ Quantum Machine Learning milestones
The following summarizes various QML projects IonQ has participated in over the past few years.
- In a partnership between IonQ and QC Ware, classical data was loaded onto quantum states to allow efficient and robust QML applications.
- On IonQ’s computers, machine learning achieved the same level of accuracy and ran faster than on classical computers.
- The project used QC Ware’s Forge Data Loader™ technology, to transform classical data onto quantum states.
- The quantum algorithm, running on IonQ’s hardware, performed at the same level as the classical algorithm, identifying the right digits 8 out of 10 times on average, the same number of times as the classical algorithm running on classical hardware.
- Researchers from IonQ and Fidelity Center for Applied Technology (FCAT) used IonQ’s cloud-based quantum computer to develop a proof-of-concept QML model to analyze numerical relationships in the daily returns of Apple and Microsoft stock from 2010 to 2018. Daily returns are the price of a stock at the daily closure compared to its price at the previous day’s closure. The metric measures daily stock performance.
- Two QML algorithms used the historical daily return data to produce a highly accurate synthetic dataset to assess the forecast accuracy. The model demonstrated that quantum computers can be used to generate correlations which cannot be efficiently reproduced by classical means, such as probability distribution.
- IonQ and Zapata Computing developed the first practical and experimental implementation of a hybrid quantum-classical QML algorithm that can generate high-resolution images of handwritten digits.
- To train the hybrid algorithm, they used NIST’s extensive database of handwritten digits.
- The results outperformed comparable classical Generative Adversarial Networks (GAN) trained on the same database. GAN is a machine learning model with two neural networks that compete against each other to produce the most accurate prediction.
1. In October 2021, IonQ became the first pure-play quantum company listed on the New York Stock Exchange.
2. While quantum computing is still in its infancy, it’s too early to select a technology that will lead to error-free quantum systems that use millions of qubits to solve world-changing problems. The technology that ultimately performs at that level may not even be in use today. Scaling to millions of logical qubits is still many years away for all gate-based quantum computers.
3. Quantum qubits are fragile and susceptible to errors caused by interaction with the environment. Error correction is a subject of serious research by almost every quantum company. It will not be possible to scale quantum computers to large numbers of qubits until a workable error correction technique is developed. I expect significant progress in 2022.
4. Technical details of the IonQ-FCAT daily stock return study are available here.
5. Technical details of the IonQ-Zapata hybrid QML research are available here.
6. Access to IonQ quantum systems is available through the cloud on Amazon Braket, Microsoft Azure, and Google Cloud and through direct API access.
Follow Paul Smith-Goodson on Twitter for current information and insights on quantum and AI