A Platoon of Green Autonomous Vehicles on Railroad Tracks
Battery-Powered Rail Cars Move Freight Cleaner, Faster, and Safer
The transportation industry’s growing emissions have sparked a race to decarbonize freight movement through innovative engineering solutions. As companies look for ways to apply cutting-edge technologies like autonomous vehicles and optimized systems, Parallel Systems, a company creating autonomous battery-electric rail vehicles, believes that adapting these advances for rail will revolutionize logistics and decarbonize transportation at the same time.
“We want to move things in a way that’s cleaner, faster, safer, and more cost-effective than how they’re moving right now,” says Jon Goh, Parallel Systems’ lead vehicle software engineer.
Founded by former SpaceX engineers, Parallel Systems aims to achieve this by combining autonomous battery-powered rail vehicle technology with a new, less expensive terminal concept and flexible operating model. While Parallel’s system is also compatible with existing terminals, the new infrastructure design allows for smaller, less expensive terminals to be built closer to shippers and customers, effectively opening new markets.
Rather than adhering to the conventional model of long freight trains, Parallel Systems’ modular approach employs a platooning system with fewer rail vehicles. This approach allows for more aerodynamic and agile movements.
The individual vehicles do not physically couple like standard rail cars. Instead, they operate in close proximity, initiating contact with buffers on the end of the rail cars to create a platoon. Once formed, each unit maintains a set pushing force against the vehicle in front. With minimal space between vehicles, aerodynamic drag decreases, and energy efficiency increases.
Parallel’s modular approach employs a platooning system where rail cars seamlessly and automatically couple and uncouple. (Video credit: Parallel Systems)
Individual rail cars split off autonomously from the main platoon, allowing freight distribution across multiple destinations from a single point to distribute deliveries. Addressing long-standing last-mile logistics challenges gives the rail industry the tools to convert some of the overburdened, $940 billion United States trucking industry to rail. Doing so with zero-emissions vehicles is a win for the environment as well.
Parallel Systems hopes to complement freight rail transportation by establishing a network of small, localized intermodal terminals because major rail yards are few and far between. This means that cargo that arrives at a rail yard needs to travel long distances by truck to its final destination. Establishing numerous smaller terminals closer to warehouses and stores enables short-haul rail services that are more economically viable, reducing transit times for crucial last-mile delivery.
“We hope to allow railroads to expand their range of routes and services and move things that normally might not move by rail,” Goh says.
Modeling the Rail Cars
Proving that autonomous rail systems operate safely and as designed is one of Parallel Systems’ most important technical priorities. Rigorous testing and validation with Simulink® is critical, according to Goh.
“Our testing agreed with the gear behavior simulated by Simscape Driveline.”
Parallel Systems relies on Simulink to evaluate control system strategies across its subsystems, such as drivetrains and brakes. It utilizes Simulink Control Design™ to develop and simulate control algorithms that must perform reliably under various operating conditions and variables, including temperature and fluid viscosities. High-fidelity simulations across a range of environmental conditions and operational scenarios enable the Parallel Systems team to robustly validate control algorithms that were first designed using reduced order models.
Simscape Driveline™ allows Parallel Systems to accurately capture phenomena such as gear backlash that would be difficult to code from scratch. The company’s engineers validate subsystem designs against performance criteria before physical implementation. The capacity for simulation of the unique dynamics involved in rail systems is invaluable for getting the engineering right from the start.
“Our job isn’t to write new backlash simulation software,” Goh says. “That’s not how we create value for our customers. Our job is to design and build vehicles and write software to control them. Our testing agreed with the gear behavior simulated by Simscape Driveline.”
Battery Modeling and Component Testing
In addition to the high-fidelity subsystem simulations in Simulink, Parallel engineers utilize MATLAB® to model battery behavior and project overall vehicle range for different operations. They’ve developed full-vehicle simulations in MATLAB to analyze how battery charge levels fluctuate based on variables such as route topography, wind conditions, and other environmental factors. MATLAB allows them to run scenarios evaluating expected battery performance and range impacts.
“We value the versatility and flexibility that MATLAB provides for our unique development process.”
When designing the platooning system, Parallel Systems procures prototype bumper damper components from manufacturers. Testing firms receive these components and generate data files containing the measured force profiles of the dampers across different speeds and displacements.
The ability of MATLAB and Signal Processing Toolbox™ to analyze diverse data file formats allows Parallel Systems engineers to import and analyze this test data. They leverage the data analysis capabilities to build lookup table models mapping the damper forces as a function of distance and speed parameters.
Parallel Systems then integrates these component damper models into Simulink simulations. This enables Parallel Systems to evaluate how the overall platooning vehicle system might perform under different scenarios and control strategies using high-fidelity models derived from the actual physical hardware data. This closed-loop workflow with MATLAB and Simulink is key to refining the platooning system design before deployment.
The bumper system that connects each autonomous rail car enables tight platooning. (Image credit: Parallel Systems)
“We value the versatility and flexibility that MATLAB provides for our unique development process,” Goh says.
Accelerating Design Iterations Through Simulation
Before building a minimum viable product (MVP), Parallel Systems used MATLAB to quickly model and simulate various vehicle configurations by adjusting variables to explore the design space and predicted performance characteristics. This upfront simulation work was critical for informing what MVP design to pursue.
“Before you spend any money, cut any steel, or buy any product, you want to put all that into a model and see how it might perform.”
With its early prototypes, Parallel Systems encountered issues such as backlash in the chain-driven axle system that impacted control performance. The subsequent design was evaluated using Simscape Driveline, which helped increase confidence in the design. Using MATLAB for modeling and data analysis capabilities was invaluable for characterizing and quantifying the backlash effect. The design team used MATLAB to improve and validate the control algorithms in their Simulink model. The team then translated these algorithm changes to code used in the physical prototype. This allowed Parallel Systems to work on improving the control algorithms on the existing prototype through software updates.
Concurrently, the company uses simulations to determine the maximum allowable backlash tolerances and vet potential redesigns to eliminate the issue in future iterations. This iterative cycle of building models and prototypes, analyzing data, updating designs and control strategies through simulations, and repeating enabled rapid learning and optimization.
Individual rail cars split off autonomously from the main platoon, allowing freight distribution across multiple destinations from a single point to distribute deliveries. (Video credit: Parallel Systems)
“Before you cut any steel or buy any product, you want to put all that into a model and see how it might perform,” Goh says.
Modeling Operational Viability
The simulation capabilities go beyond evaluating vehicle designs and performance characteristics. Parallel Systems engineers use MATLAB and Simulink to simulate potential business scenarios and financial implications. By modeling variables such as vehicle mass, costs, routes, speeds, and revenues, they can estimate operating expenses and profitability projections and analyze the overall viability of their concept across different configurations. This ability to rapidly run simulations evaluating many different design points and operating scenarios is critical.
“Simulink simulations allow us to vet whether our freight rail plans are economically feasible before significant investment. The simulations provide evidence that, with the right vehicle parameters, we can achieve profitable operations at scale while making a major impact on freight transportation.”
“Simulink simulations allow us to vet whether our freight rail plans are economically feasible before significant investment,” says Goh. “The simulations provide evidence that, with the right vehicle parameters, we can achieve profitable operations at scale while making a major impact on freight transportation.”
Validating the autonomous piece is also critical to the company’s vision. Parallel Systems engineers are developing high-fidelity maps of rail tracks with GPS data for initial, coarse localization of their autonomous trains. They leverage MATLAB to run simulations on recorded test data sets incorporating GPS. This allows them to rapidly iterate on algorithm designs by evaluating performance on real data.
They also plan to incorporate vision-based localization techniques, taking advantage of the fixed rail network. Trains on a fixed network simplify the localization problem compared to autonomous cars on open roads. A key advantage is that the safe state for a train is well-defined—just stop the train. Even with a more straightforward issue to solve than self-driving cars, Parallel Systems rigorously validates its autonomous systems before scaling up deployment, relying on advanced braking systems to ensure safety.
Future Projects
While making strides with its novel freight rail concept, Parallel Systems is investigating building more of the rail car components rather than integrating off-the-shelf parts. This includes developing custom electric motor designs optimized for autonomous rail applications. With this approach, MATLAB and Simulink for new modeling and validation needs are critical.
Parallel Systems also plans to adopt Model-Based Design for additional design, which will extend to safety-critical systems development for next-generation vehicles.
Simulink capabilities for multidomain system modeling enable comprehensive virtual testing of motor designs along with other tightly coupled elements like drivetrains and control systems. Toolboxes such as Simscape™ aid in analyzing the detailed physics and dynamics involved.
Parallel Systems also plans to adopt Model-Based Design for additional design, which will extend to safety-critical systems development for next-generation vehicles. By tying requirements and testing back to system modeling, Parallel Systems aims to validate robustness and reliability more rigorously from the earliest design stages through an integrated, simulation-driven workflow.
“A lot of what we’ve done is learning how to build a train,” Goh says. “The next phase of the company is to build a product that can do regular service, meets customer requirements, and can be verified from a regulatory and safety standpoint.”
Read Other Stories
AEROSPACE
Pilotless Helicopters Deliver Large Payloads with Precision
Autonomous Helicopter Startup Tests Prototype for Drone Deliveries
AI / ROBOTICS
Autonomous Electric Tractor Brings Artificial Intelligence to the Field
California Vineyard Finds Driver-Optional Smart Vehicle Has Power to Spare
GREEN TECHNOLOGY / CONTROL SYSTEMS
Hydrogen Fuel Cells Reduce CO2 Emissions
Electrifying Commercial Vehicles with Hydrogen Fuel Cells
