By Jon Crowley
The latest BRP sway bar patent application raises an interesting question: What if a UTV could look at the terrain ahead and choose its sway bar setting before reaching it?
Published October 8, 2026, the application describes an automatic anti-roll bar control system that uses camera images and a trained machine-learning model to select operating modes. The idea is to anticipate upcoming conditions instead of relying entirely on what the vehicle is already doing.
That makes it particularly relevant as electronically controlled sway bars reach production sport UTVs. We have tested the Polaris RZR Pro R Ultra with SDi-developed active sway bar links and the Kawasaki TERYX H2 Limited eS with FOX electronic sway bar control. Both demonstrate why freeing the suspension over uneven ground, then adding roll support when needed, can improve the driving experience.
However, our testing also leads to another question: Would a forward-facing camera deliver a bigger benefit to the shocks than to the sway bars?
What BRP Filed
The application is titled Automatic Anti-Roll Bar Control System. Its applicant is Bombardier Recreational Products Inc., and the listed inventors are William Lamey and Jeffrey Best.
- Publication number: US 2026/0307448 A1
- Application number: 19/173,162
- Filed: April 8, 2025
- Published: October 8, 2026
This is a published patent application, not a production announcement or a granted patent. It does not identify a specific future Can-Am model, release date or retail package. The description expressly includes side-by-sides and UTVs among its possible vehicle platforms, but its scope extends beyond those vehicles.
How Camera-Based Sway Bar Control Would Work
The core process starts with a vehicle-mounted camera capturing images in the direction of travel. A trained model processes those images and predicts suitable anti-roll bar modes. A separate decision stage then determines which mode to implement.
One example uses a forward-facing, two-dimensional visible-light camera. The system can resize and crop the images before processing them, reducing the computing resources needed. The description gives an example of anticipating terrain the vehicle will encounter within the next one to three seconds.
In practical terms, the objective is to recognize upcoming conditions early enough to prepare the sway bar system. That could mean allowing more independent wheel movement through rough terrain or increasing the connection between the left and right suspension when greater roll support is appropriate.
The application describes minimum- and maximum-stiffness modes, with optional intermediate settings. It also discusses both discrete mode selection and an alternative that provides continuously varying outputs. These are disclosed examples, not specifications for a confirmed production system.
The Drawings Show Two Hardware Approaches
Figure 1B illustrates controllable damper links that change how strongly the sway bar couples the suspension. A valve controls fluid movement within the link, allowing its resistance to change. The description accommodates configurations with one electronically controlled link or links at both ends.
That distinction matters because changing the link’s behavior changes the effective response of the sway bar assembly. It does not require changing the torsional stiffness of the steel bar itself.
Figure 1C shows another approach: a mechanical coupling that connects or releases sections of the bar. In the described example, locking the coupling provides maximum roll resistance, while unlocking it permits greater independent movement.
The filing therefore addresses automatic control across more than one hardware arrangement. Its central idea is how the system chooses a mode from camera information, rather than a commitment to one particular sway bar mechanism.
It Does More Than React to a Single Image
The application describes filtering a sequence of predicted modes before acting. It also includes checks on elapsed time since the last change, permissible transitions and compatibility with the vehicle’s operating condition.
Vehicle-state information can include speed, steering angle, pitch, yaw, roll, braking, throttle and the current suspension state. Depending on the implementation, those inputs help determine whether a predicted mode is appropriate to apply.
Those details are important. A system that changes its mind with every camera frame could produce inconsistent behavior. BRP describes ways to stabilize the decision and control how frequently modes change.
It is also worth distinguishing the rate of image processing from the rate of actual mode changes. The filing describes examples with frequent predictions but slower discrete mode updates, as well as faster alternatives. It should not be read as a promise to adjust the links separately for every individual bump.
Learning From an Expert Driver
One disclosed training method records an expert driver manually selecting modes while a forward-facing camera records the terrain. The mode selections and images receive timestamps so they can be matched into labeled training examples.
The model then learns associations between what the camera sees and the operating mode selected by the driver. The description also discusses reviewing the data and removing unsuitable images, including frames obscured by debris or a passing branch.
This is a specific example of machine learning applied to vehicle setup. It does not mean the vehicle necessarily learns a new suspension calibration from each owner during normal use.
Polaris and SDi: What We Have Already Tested
The Polaris RZR Pro R Ultra’s DYNAMIX DVS system combines electronically controlled shock damping with SDi-developed hydraulic locking sway bar links. The driver can select Active, Locked or Unlocked settings.
In our Active-versus-Locked comparison, Active felt more comfortable over uneven ground while retaining similar perceived cornering support. In the rocks, the Ultra also felt more planted. Our crawling vehicle had larger sticky tires and other changes, so the links do not deserve sole credit for every improvement.
During our technical discussion with Polaris and SDi, they explained that the links normally remain unlocked in the semi-active setting and engage as cornering demands increase. That approach closely matches what I want from a sway bar: wheel independence over rough ground, with roll support when turning.
Kawasaki and FOX: Another Production Example
The two-seat 2027 Kawasaki TERYX H2 Limited eS combines FOX Live Valve Internal Bypass shocks with electronic sway bar control through KECS+.
Kawasaki describes switching between Lock and Flex conditions while coordinating shock damping and sway bar effectiveness. Our first test concentrated on Double Sammy at Sand Hollow, primarily in Comfort mode. The stock machine articulated well, and the sway bar transitions were difficult to feel.
The shock and sway bar behavior are tied to the selected suspension mode. That is a different driver-control arrangement from the separate sway bar selections available on the Polaris Ultra.
FOX Has Already Disclosed Camera Inputs
Camera inputs also appear in FOX’s earlier US 11,634,003 B2 patent, Automated Control System for an Electronically Controlled Sway Bar Link, granted April 25, 2023.
That document lists single- and multispectrum cameras among possible sensor inputs. Its description discusses automatic adjustment of suspension dampers, sway bar damper links and physical disconnects using sensor and environmental information. Camera inputs also appear in claims 11 and 15.
BRP’s application develops a particular camera-image and trained-model mode-selection approach. The presence of cameras in both documents does not, by itself, establish that they cover the same invention. It also does not demonstrate that a camera-based system is installed on either production UTV discussed here.
Why I See a Bigger Opportunity for Predictive Shock Control
Based on my testing, I generally prefer the sway bars to remain released over uneven terrain until cornering calls for additional support. That preference comes from the improvement in wheel independence and ride comfort I have felt with electronically controlled links.
A forward-facing camera could help recognize a terrain transition before the tires reach it. However, I think its more compelling application may be preparing the shocks for an upcoming bump, compression or G-out.
The shocks manage how quickly the suspension compresses and rebounds. Seeing an approaching event could give a controller extra time to choose appropriate damping before the suspension begins moving through it.
A sway bar primarily adds resistance to differences in left- and right-side suspension movement. When both wheels on an axle move together by similar amounts, the bar contributes relatively little torsional resistance. An offset impact or diagonal ditch is different because the wheels move at different times or by different amounts.
That is why I find BRP’s emphasis on anti-roll bar control interesting. For the bumps and G-outs I encounter, predictive shock adjustment seems likely to offer the more direct benefit. This is my assessment from driving, not a measured result from BRP’s proposed system.
There are still plausible benefits for sway bar prediction. An upcoming off-camber section, uneven corner or transition into rough terrain could affect the best balance between compliance and roll support. A useful system would need to combine that visual information with speed, steering and vehicle motion, since the camera cannot know the driver’s intended path from terrain appearance alone.
What Would Make This Useful on a UTV?
For a future camera-based system, I would want to know how it handles dust, changing light and an obstructed lens; how it responds when the driver changes line; and what it does when visual predictions conflict with current vehicle motion.
I would also want back-to-back testing with the predictive layer enabled and disabled. The production systems we have driven already provide useful comfort and articulation without a disclosed forward-looking camera control strategy. Additional complexity needs to produce an improvement a driver can feel.
The BRP sway bar patent application provides a detailed look at one possible direction for suspension control. Polaris/SDi and Kawasaki/FOX have already demonstrated the value of electronically managing the sway bar connection. The next question is how much more can be gained by seeing the terrain before the vehicle reaches it—and which suspension components benefit most.





















