Battery Optimized Onsite Swapping Technology:
BOOST
BOOST System Overview Video
Battery Optimized Onsite Swapping Technology, or BOOST, is a robotic architecture designed to address the problem of finite battery life in field robotics. Current mobile robots such as Boston Dynamics' 'Spot' and i-Robot's Roomba are unable to continuously operate and must sit on a charging pad for 50% of their lifetime. With BOOST, robotic downtime is completely eliminated, as minibots can quickly swap their battery going from low to 100% in about 90 seconds.
This project was developed as part of my Senior Capstone design course part of a group of 10 engineers. Together we designed and manufactured two minibots and one hub robot for about $6000, completing rigorous numerical analysis and validation of our system.
After capstone, we developed the system further. Conducting additional validation and trade studies to produce a publication for IEEE Aerospace 2024 entitled "Battery-Swapping Multi-Agent System for Sustained Operation of Large Planetary Fleets".
Hub - Larger robot with onboard power generation to sustain minibot swarm.
Minibot - Small modular robot outfitted for task completion.
System Overview
We developed a robotic architecture where you have a hub robot that is outfitted with onboard power generation which sustains a swarm of mini robots in the field. The hub robot houses a cashe of batteries, and a battery swap mechanism which reliably picks up the minibot on uneven terrain, and swaps a spent battery for a fresh one. During the swap, electrical continuity with the minibot is maintained to avoid shutting down the robot to further decrease robotic downtime. The allocation of power generation to the hub enables the minibots to be outfitted for specific tasks, and although our system only operates with two minibots, the architecture is easily tile-able and well suited for scaling to a larger fleet.
One of the most novel aspects to our project is the robots' ability to dock and carry out battery exchange on uneven terrain. Existing robotic battery swap mechanisms have only been demonstrated on level terrain, so our project aimed to optimize our system to expand the use cases for battery swap to more dynamic environments. As a mechanical group, we focused our efforts on expanding the range of misalignment the system can compensate mechanically through geometry optimization.
Docking Geometry Optimization
In order to rapidly test and iterate upon hardware to ensure reliable docking, a rigid body physics simulation was developed using MATLAB Simscape Multibody. First a simulation developed to evaluate the axial displacement limits of various lifting wings through a custom "axial iterative solver" algorithm, which uses a binary search method to rapidly find misalignment bounds within a time complexity of 𝑂(𝑙𝑜𝑔(𝑛)). This drastically cut our iteration time and enabled the robust qualification of lifting geometry design.
Lift arms and lifting MATLAB Simscape simulation
The team built an additional Simscape physics environment to determine the validate hub and minibot bumper geometries for entry into the battery swap mechanism. To accelerate the process of finding the optimal bumper geometry, a geometric solver was developed, which performed a grid search over a parameterized curve to find the highest-performing bezier curve which was implemented on the final robot.
Entry optimization grid search results and implemented optimal geometry
Monte Carlo testing results of initial design vs computationally optimized design.
The team further optimized the entry geometry with the addition of bumpers on the front of the minibot. implemented Monte Carlo testing to evaluate minibot bumper geometry. Testing hundreds of different starting orientations, our simulation defined a region of failures and successes, the volume of this region was used as a score mechanical compensation. Using this testing method to iterate, the optimal bumper design led to an increase in success volume of of 3.6x!
Validation
To ensure our simulation matched with reality, we conducted a final physical test campaign using to characterize the limits of our system. A motion tracking script using Canny edge detection algorithms was written in MATLAB to track the position of the minibot driving from random start positions. The simulation region did match was the physical results, and the final system could compensate for ± 20.11 ̊
of yaw misalignment.
(Top) MATLAB script tracking minibot position for overhead testing. (Bottom) Overall Docking Success Region
Outcome and Future Work:
Thanks to our team's hard work, we were awarded the Gorlov Innovation Award an annual award given to mechanical engineering students at Northeastern University who have demonstrated exceptional creativity and innovation in their capstone projects. The award selection committee evaluates each capstone project based on several criteria, including the degree of innovation, the impact of the project on society, the quality of the design and analysis, and the feasibility of the project's implementation. The committee also takes into consideration the student's ability to communicate their ideas effectively.
We believe that our system, BOOST, can revolutionize the field of mobile robotics as it directly addresses one of the most limiting challenges facing the field. As a result, our team is continuing to work with Northeastern's RIVeR Lab to develop it into more meaningful work.
(Left) MechE Team: Track winners and Gorlov Innovation Award Winners. (Right) BOOST Team on Mechanical Engineering Capstone Day.