07/08/2026
By Manav Shaileshkumar Mistry

The Kennedy College of Sciences, Miner School of Computer and Information Sciences, invites you to attend a Master’s thesis defense by Manav Shaileshkumar Mistry on “Performance Evaluation of Quadruped Locomotion Under Varied Payload and Operating Conditions: Experimental Validation on Boston Dynamics Spot".

Candidate Name: Manav Shaileshkumar Mistry
Degree: Master’s
Defense Date: Monday, July 20, 2026
Time: from 10 to 11 a.m.
Location: 309 Danedeneau, North Campus

Thesis/Dissertation Title: Performance Evaluation of Quadruped Locomotion Under Varied Payload and Operating Conditions: Experimental Validation on Boston Dynamics Spot

Committee:

  • Advisor: Reza Azadeh, Ph.D., Associate Professor, UMass Lowell
  • Maria E. Cabrera, Ph.D., Assistant Professor, UMass Lowell
  • Samantha Reig, Ph.D., Assistant Professor, UMass Lowell

Brief Abstract:

Existing research has made substantial progress in quadruped motion planning and control, yet the impact of payload distribution on locomotion performance remains under-investigated. Previous work on payload-carrying quadrupeds has primarily focused on online payload identification and adaptive control, while other studies have increased load-carrying capacity through novel mechanical designs. However, these efforts have not systematically examined how different payload distributions affect locomotion performance across a range of operating conditions.

To systematically evaluate payload distribution in quadruped locomotion, this thesis makes three primary contributions. First, we developed a systematic experimental framework featuring reproducible test terrains with controlled difficulty levels and a custom-designed payload saddle for repeatable payload-carrying evaluations. Second, we collected a dataset comprising 93 controlled experiments conducted using the Boston Dynamics Spot quadruped. The dataset includes high-frequency measurements from the robot's onboard sensors, including joint torques, joint velocities, and inertial measurement unit (IMU) data. Third, we analyzed the data at both the experiment level (unsegmented) and the stride level (segmented) using metrics that quantify energy efficiency, front-rear load symmetry, and gait consistency.

Our evaluation compared locomotion performance across multiple conditions, including payload distributions, terrain types, payload weights, and payload types (static versus dynamic). We also compared Spot's payload-aware locomotion mode with its default payload-unaware controller.

The results demonstrate that payload placement significantly affects energy consumption, joint loading patterns, and gait consistency. Rear-weighted configurations consistently reduce performance, whereas centrally distributed payloads improve front-rear load balance and gait stability. Analysis of the experiment-level (unsegmented) data further revealed a linear relationship between payload position and front-rear load asymmetry across front-, center-, and rear-mounted configurations. Linear regression predicted that front-rear load symmetry (zero load ratio) is achieved with the payload positioned slightly forward of the robot's geometric center, and subsequent experimental results validated this prediction.