Electrification and platooning are two major technology drivers that can lead to both lower costs and emissions for transportation. The on-going project continues to identify the synergy between these two mega-trends around platooning and electrification. In the first year of the project, we have built high-fidelity computational fluid dynamics models to estimate the effects of platooning on the aerodynamics and subsequently, the energy consumption. In the second year of the project, we have built machine-learning accelerated models to determine the aerodynamics of platoons. In this project, we will now be able to address several key questions around (i) the effect of platooning configurations and added sensor stack for platooning and (ii) trade-off between safety and energy savings related to velocity smoothing.
At present, about 51 percent of freight miles in the U.S. are in states that approve commercial truck platooning. This provides a technology that can allow drivers to link their vehicles’ active safety systems to enable vehicle following at an aerodynamic following distance. Our industrial partner, Peloton is a connected and automated vehicle technology company dedicated to improving the safety and efficiency of U.S. and global freight transportation, which is backed by ten Fortune Global 500 companies. Peloton’s flagship driver-assistive platooning system links the active safety systems of pairs of trucks and connects them to a cloud-based Network Operations Cloud (NOC) that limits platooning to appropriate roads and conditions. Peloton solutions also require best-in-class forward collision avoidance systems and other safety features, thereby incentivizing their adoption.
In the first year of this project (#179), the team has built a close tight-knit collaboration with the CMU team's expertise in electrification complemented by Peleton's experience on platooning. We have built CFD capability utilizing the PI's high-performance computing infrastructure as well as CMU's ANSYS partnership that allows exploration of platooning configurations and its effect on drag reduction.
In the second year of this project (#285), the team built machine-learning accelerated surrogate models based on convolutional neural net (CNN) to determine the flow field and aerodynamics given the vehicle geometry.
In the current year, the following research goals will be carried out with the following tasks:
Task 1: Quantify the Energy Consumption including sensor stack and computing
1.1. Aerodynamics and Energy Consumption for various sensor stack
Using the machine-learning accelerated CFD surrogate models based on CNNs, we will predict the aerodynamics and the energy consumption due to the sensor stack needed to ensure platooning. This effort will be using the insights provided by Peloton, our industrial partner.
1.2. Computing Energy Budget
Develop a detailed computing energy budget for the sensor stack required for platooning and integrate this into a full vehicle dynamics model. The computing energy will include energy cost associated with any additional computing hardware and energy cost of sensor.
Task 2: Trade-off between Safety vs Velocity Smoothing
2.1 Velocity smoothing vs Safety trade-off
The following vehicle in the platoon can drive smoother which can lead to energy savings. The extent of this energy saving is determined by the degree of smoothening of the velocity profile. However, this comes at the cost of safety as the following distance is now not constant. We have identified a trade-off between these two quantities and we will explore this trade-off in detail.
2.2 Comparison to real-world smoothing:
The velocity smoothing identified will be compared against real-world profiles with input from Peloton Tech.
Task 3: Deployment and Validation
The tasks will seek input from our very active collaboration with the deployment partner, Peloton Technologies and we will continue to perform validation effort. In particular, we will perform validation against real-world performance.
 A. Mohan, S. Sripad, P. Vaishnav, V. Viswanathan, arXiv:1908.08920 [cs.CY], Nature Energy (under revision)
Task 1: Quantify the Energy Consumption including sensor stack and computing Q1-Q2
1.1. Aerodynamics and Energy Consumption for various sensor stack -- Q1-Q2
1.2. Computing Energy Budget -- Q2
Task 2: Trade-off between Safety vs Velocity Smoothing Q3-Q4
2.1 Velocity smoothing vs Safety trade-off - Q3-Q4
2.2 Comparison to real-world smoothing: Q4
Task 3: Deployment and Validation: Q1-Q4
The project leverages our continued collaboration with Peloton. Peloton has ongoing use-case projects in several geographical areas. As of January 2019, Peloton Technology’s platooning demonstration project in Florida completed over 1,000 miles. Furthermore, over nine states in the United States, namely, Arkansas, Georgia, Michigan, Nevada, North Carolina, Ohio, South Carolina, Tennessee and Texas have permitted commercial deployment of platooning for Class 8 semi-trucks. Peloton Technology also has ongoing projects under the Federal Highway Administration (FHWA) and the US Department of Transportation (USDOT).
The computational and theoretical analysis undertaken by the PI is being validated at various stages with data directly from Peloton. The results will enable a clear map of platooning and sensor, computing effects for the energy efficiency of semi-trucks.
Expected Accomplishments and Metrics
Similar to that described in project #179 and #285, two distinct tangible outcomes from the computation front are:
1. High-fidelity energy-efficiency maps for platoons
2. A safety-energy reduction to velocity smoothing for the platooning vehicle.
On the deployment front:
1. Validation of velocity smoothing results using Peloton testing.
2. The longer-term goal will be to deploy and demonstrate results for groups/fleets of electric vs diesel semi-trucks.
||Student - PhD
||Student - PhD
||Faculty - Untenured, Tenure Track
Amount of UTC Funds Awarded
Total Project Budget (from all funding sources)
|Data Management Plan
||Dec. 5, 2019, 5:25 p.m.
||methodologies to solve spatiotemporal dynamical systems using convolutional neural ODEs
||Sept. 30, 2020, 7:04 p.m.
||Sept. 30, 2020, 7:05 p.m.
||Performance metrics required of next-generation batteries to make a practical electric semi truck
||Dec. 21, 2020, 11:51 p.m.
||Evaluating the potential of platooning in lowering the required performance metrics of li-ion batteries to enable practical electric semi-trucks
||Dec. 21, 2020, 11:52 p.m.
||Performance metrics required of next-generation batteries to electrify vertical takeoff and landing (VTOL) aircraft
||Dec. 21, 2020, 11:52 p.m.
||Quantifying the Economic Case for Electric Semi-Trucks
||Dec. 21, 2020, 11:53 p.m.
||Potential for electric aircraft. Nature Sustainability
||Dec. 21, 2020, 11:54 p.m.
||The Future of Vehicle Electrification in India May Ride on Two Wheels
||Dec. 21, 2020, 11:55 p.m.
||Performance metrics required of next-generation batteries to electrify commercial aircraft
||Dec. 21, 2020, 11:56 p.m.
||Trade-offs between automation and light vehicle electrification
||Dec. 21, 2020, 11:56 p.m.
||Universal Battery Performance and Degradation Model for Electric Aircraft
||Dec. 21, 2020, 11:57 p.m.
||April 10, 2021, 11:21 a.m.
No match sources!
||Deployment Partner Deployment Partner