AutoPeel: Adhesion-aware Safe Peeling Trajectory Optimization for Robotic Wound Care

TL;DR

The study addresses automated, adhesion-aware wound dressing removal to reduce pain and healthcare costs in chronic wounds. It introduces AutoPeel, an XPBD-based multi-softbody adhesion model coupled with gradient-based MPC to optimize safe peeling trajectories along the wound surface. A peeling objective uses adhesion boundary layers V1 and V2 with a gamma parameter to encourage detachment of the immediate boundary while limiting deformation in deeper layers; the MPC minimizes a loss $L = \mathcal{H}(\mathbf{x}_{t+h}) + \alpha \sum_{t=t_0}^{t+h} \mathcal{P}(u_t) - \beta \cdot \mathcal{S}(A_t, A_{t-1})$ to balance safety and progress. Experiments on skin phantoms and a real human demo show reduced skin deformation and feasible transfer from simulation to practice, highlighting potential for automated safe wound care.

Abstract

Chronic wounds, including diabetic ulcers, pressure ulcers, and ulcers secondary to venous hypertension, affects more than 6.5 million patients and a yearly cost of more than $25 billion in the United States alone. Chronic wound treatment is currently a manual process, and we envision a future where robotics and automation will aid in this treatment to reduce cost and improve patient care. In this work, we present the development of the first robotic system for wound dressing removal which is reported to be the worst aspect of living with chronic wounds. Our method leverages differentiable physics-based simulation to perform gradient-based Model Predictive Control (MPC) for optimized trajectory planning. By integrating fracture mechanics of adhesion, we are able to model the peeling effect inherent to dressing adhesion. The system is further guided by carefully designed objective functions that promote both efficient and safe control, reducing the risk of tissue damage. We validated the efficacy of our approach through a series of experiments conducted on both synthetic skin phantoms and real human subjects. Our results demonstrate the system's ability to achieve precise and safe dressing removal trajectories, offering a promising solution for automating this essential healthcare procedure.

Figures (7)

• Figure 1: An illustration of the robotic wound dressing change concept presented in this paper. A safe, efficient dressing peeling trajectory is first optimized in a simulated environment and then transferred into a real-world scenario.
• Figure 2: Left: adhesion modeling and fracture mechanism of adhesive connections. Adhesion is modeled with distance constraints between elements in $\mathbf{x}^{s}_t$ and $\mathbf{x}^{d}_t$. An adhesion constraint is fractured when its potential energy $E^{i, j}$ exceeds a limit $\varepsilon$. Right: visualization of current removed connections $V^{\text{det}}_t$, the first immediate adhesion layer $V^1_t$, and the second immediate adhesion layer $V^2_t$.
• Figure 3: A visualization of our procedures for finding the adhesion boundary. From figure (a)-(f), finds the 1st immediate adhesion boundary layer $V^1_t = \mathcal{D}(V^{\text{det}}_t)$. To further obtain the 2nd adhesion boundary layer, we apply the procedure recursively to get $V^2_t=\mathcal{D}(V^{\text{det}}_t \cup V_t^1)$.
• Figure 4: Our physical setup where the controlled experiments' results were collected. The wound dressing (in orange) is adhered to a foam phantom (in blue). The RGBD camera sees the bottom side of the phantom. There are 6x8 green landmarks labeled on the back of the foam. We evaluate the deformation of the phantom by combining depth and landmark tracking from the RGB image. An example point cloud output from the camera is shown on the right.
• Figure 5: Comparison of mean and standard deviation of landmarks displacement results overtime of different methods. The plot has shown that our method causes the lowest average landmark displacement overall, showing that the method can achieve the peeling goal in a safer manner.