The research, led by Ankit Nayak of the University of Petroleum and Energy Studies in India together with Yu-Xiong Su and Jane Jingya Pu of the University of Hong Kong, builds on computer-assisted surgery (CAS), the current standard of care for jaw reconstruction. CAS combines three-dimensional virtual surgical planning with computer-aided design and manufacturing, and in cases of segmental mandibulectomy—the surgical removal of a portion of the lower jaw—the well-established reconstructive approach remains the vascularized free fibula flap. Patient-specific cutting guides and implants transfer the virtual plan to the operating theater, reducing freehand contouring and repeated intraoperative measurement while maintaining the planned three-dimensional position of the fibular segments relative to the residual mandible.

But geometric accuracy alone does not guarantee mechanical safety. Uniform-thickness, uniform-width plates embody two competing design risks. Plates that are overly thick have been associated with stress shielding, which may impair bone healing, increase the risk of implant exposure through overlying soft tissue, and even contribute to radiation scattering during post-operative radiotherapy. Plates that are too thin carry a higher risk of fracture, which in turn has been linked to bone non-union or mal-union in some reports. Compounding the problem, stress in a reconstruction plate accumulates disproportionately near sharp edges, notches, and screw holes—meaning a single uniform cross-section may be structurally inefficient by design.

The team’s solution is a parametric-design algorithm that feeds finite element analysis (FEA) results directly back into the geometry-generation process. Rather than uniformly reinforcing the entire plate, the algorithm selectively thickens only the regions where von Mises stress concentrations exceed a defined threshold. Crucially, the reinforcement is blended smoothly into the plate’s geometry using a cubic smoothstep taper that provides C1-continuous transitions—meaning the width changes gradually with zero slope at both boundaries. This prevents the abrupt changes in cross-section that would otherwise create new stress concentrations, the very problem the reinforcement is meant to solve.

The mathematical machinery behind the adaptation is elegant in its economy. The plate is represented as a sequence of rectangular cross-sections swept along a three-dimensional centerline guide curve, each rectangle defined in its own local orthonormal frame derived from the actual vertex coordinates. When a zone is flagged for reinforcement, the algorithm scales the width of the rectangles within that zone while holding thickness constant—a constraint that reflects the surgical requirement for a plate of known, uniform thickness that can be pre-bent intra-operatively. Scaling is performed about each section’s centroid, ensuring the plate centerline and its load path along the mandible are never displaced, and the counter-clockwise vertex ordering is preserved so the downstream surface triangulation produces a valid, non-self-intersecting mesh.

To validate the approach, the researchers performed FEA on a baseline plate simulating right molar clenching, with muscle forces adopted from published literature and the bone modeled as dense cortical bone with a Young’s modulus of 13,700 MPa. The screws were simplified as cylindrical shafts with bonded contacts, and a mesh convergence study confirmed numerical reliability—refining the mesh from 127,947 to 144,521 nodes changed the peak equivalent stress by only 0.39%. The stress-versus-arc-length profile revealed two zones where stress remained at or above a threshold of 100 MPa, roughly one-third of the baseline global maximum of 305.01 MPa: Zone 1, between screw holes 2 and 4, and Zone 2, between holes 8 and 10.

The results demonstrate the power of targeted reinforcement. Plate A, which widened only Zone 1 to 6 mm, reduced peak stress there by 36.0%—down to 195.2 MPa—and strain by 43.8%, the largest reduction observed across both design variants. Remarkably, even the unmodified Zone 2 benefited, with stress and strain falling by 12.7% and 14.3% respectively, indicating that local width adaptations help redistribute load across the entire plate. Plate B, which reinforced both zones, pushed Zone 2 stress down by 30.6% and strain by 33.3% relative to baseline, while Zone 1 strain dropped by approximately 40.6%.

One finding deserves particular attention: Plate B showed a slightly higher stress at Zone 1 than Plate A—204.64 MPa versus 195.20 MPa. The authors interpret this as a non-local mechanical trade-off rather than a design failure. Reinforcing Zone 2 alters the plate’s global deflection and curvature profile, which increases the local bending moment, and hence stress, at Zone 1. Nevertheless, Zone 1 stress in Plate B remained 32.9% lower than the baseline value, demonstrating that simultaneous multi-region adaptation can achieve favorable stress reduction at both critical locations without reintroducing baseline-level concentrations at either site.

The study also situates its approach against topology optimization, the alternative computational strategy that iteratively redistributes material within a design domain using sensitivity-based solvers. Topology-optimized implants have been reported with better fatigue performance, but they typically require repeated iterations between optimization and FEA, along with post-processing to correct numerical instabilities and satisfy manufacturing constraints. The proposed adaptive-scaling method, by contrast, recommends manufacturable, screw-hole-constrained geometry in a single FEA-driven feedback loop, offering a predictable design path at the expense of not exploiting the full unconstrained material-distribution space that topology optimization provides. The authors note that a direct benchmarking of computational cost was not performed and remains a task for future work.

The researchers are candid about the study’s limitations. The evaluation rests on a single patient-specific baseline reconstruction under a single static loading condition, with muscle forces drawn from published sources and screws idealized as bonded cylindrical shafts. No physical prototyping, mechanical testing, or fatigue analysis was conducted, even though reconstruction plates in vivo experience cyclic masticatory loading. The algorithm currently handles only single-barrel reconstruction cases and cannot accommodate branched implant structures. Importantly, the team emphasizes that automation alone will not guarantee better clinical outcomes: key design parameters remain surgeon-specific, allowing clinicians to define them according to their unique requirements—a deliberate balance between mechanical performance and surgical objectives. With the baseline plate’s peak stress of 305.01 MPa sitting well below the yield strength of 3D-printed Ti-6Al-4V titanium alloy—524.2 MPa in the lowest case of electron beam melting and up to 1035.7 MPa with selective laser melting—the approach offers a computationally efficient pathway toward implants that are not only anatomically tailored but mechanically optimized for the loads they will actually encounter.

Subject of Research: Stress-informed parametric design of patient-specific titanium implants for computer-assisted jaw reconstruction

Article Title: Stress-informed parametric shape adaptive patient-specific implant design in computer-assisted jaw reconstruction

Article References: Nayak, A., Su, Y.-X., & Pu, J. J. (2026). Stress-informed parametric shape adaptive patient-specific implant design in computer-assisted jaw reconstruction. Frontiers in Oncology, 16, Article 1966850. https://doi.org/10.3389/fonc.2026.1966850

Image Credits: AI Generated

DOI: 10.3389/fonc.2026.1966850

Keywords: patient-specific implants, mandibular reconstruction, finite element analysis, 3D printing, titanium alloy, stress adaptation, computer-assisted surgery, parametric design, von Mises stress, fibula free flap, biomechanics, additive manufacturing

The study, led by Kaitao Ren and corresponding author Zhigui Ren, introduces what the authors call a Structure-Mechanism Collaborative Optimization approach for the stick of a 20-ton hydraulic excavator. The central insight is that mechanism parameters—the geometry of hinge points and linkage dimensions that govern how hydraulic cylinders convert their motion into digging force—and structural parameters such as plate thicknesses are deeply coupled. Most previous studies have optimized these two aspects separately, adjusting either the linkage geometry or the steel plate gauges without accounting for how a change in one domain alters the loads and stresses in the other. By explicitly modeling that coupling, the team argues, designers can find solutions that neither approach could reach alone.

The researchers began with physical evidence rather than pure simulation. Using experimental data from the machine, they calculated the stick’s digging force and the equivalent von Mises stress—a standard measure that combines the multi-directional stress state at a point into a single comparable value—across the working envelope. From the measured stress distributions they identified the critical dangerous conditions, the specific postures and load cases in which the stick is most at risk of failure. Anchoring the optimization in test-identified hazardous conditions is a key methodological choice, because it ensures the design effort targets the scenarios that actually break hardware in the field rather than idealized load cases that rarely occur.

With those conditions established, the team confronted a familiar computational bottleneck. Evaluating digging force and stress across the full design space of hinge geometries and plate thicknesses would require thousands of finite element analyses, each computationally expensive. Their solution was to combine Optimal Latin Hypercube Design, a sampling strategy that spreads test points efficiently across a multi-dimensional space, with Kriging surrogate models—a class of statistical interpolators, originally developed in geostatistics, that can approximate the output of expensive simulations with quantified accuracy. Three surrogate models were built, one each for digging force, maximum equivalent stress, and stick mass, allowing the optimization algorithms to explore the design space at a fraction of the computational cost of repeated full simulations.

The optimization itself was organized as a two-level Collaborative Optimization model, a multidisciplinary design framework in which individual disciplines are optimized in parallel under coordination from a system-level optimizer, with consistency among the disciplines enforced through shared targets. The model was solved using intelligent optimization algorithms, including NSGA-II, a widely used evolutionary algorithm well suited to problems with multiple competing objectives, under practical engineering constraints. The objectives pull in different directions: more digging force generally means heavier structure or higher stress, while reducing weight tends to raise stress levels. The team weighted the objectives empirically, assigning higher priority to digging capacity and structural safety—regarded as the primary requirements for hard-soil excavation—and a lower weight to mass reduction as a secondary goal.

The reported results are striking in their balance. The optimized design increases theoretical digging force by 9 percent, reduces the maximum equivalent stress in the stick by 7 percent, and cuts the stick’s weight by 9 percent. Achieving all three simultaneously is the noteworthy part: typically, a designer must trade one against another. A stronger, lighter arm that is also less stressed represents a genuine Pareto improvement, meaning the collaborative optimization found a design that dominates the baseline on every metric considered. For manufacturers, a 9 percent weight reduction translates directly into material savings and reduced machine mass, while the stress reduction promises longer component life under the cyclic loading that defines hard-soil work.

Verification of the surrogate-based results was addressed through recalculation. When the optimized design variables were substituted back into the digging force calculation and stress simulation models previously developed and validated by the research group, the deviations between surrogate predictions and the finite element verification values were 1.368 percent for digging force, 1.144 percent for maximum equivalent stress, and 1.414 percent for stick weight—all within 1.5 percent. According to the authors, these small discrepancies confirm that the Kriging surrogate models possess sufficient approximation accuracy and that the collaborative optimization results are physically reliable and applicable in engineering practice. The team has acknowledged that limited funding prevented further physical experimental validation at this stage, a point they intend to elaborate more clearly in the revised manuscript.

One technical detail deserves attention from anyone who has modeled pinned machine structures: the load scheme. Rather than applying idealized point loads at the hinge bores, which are known to produce spurious stress concentrations that do not exist in real pin connections, the researchers adopted a cosine-distributed loading scheme to reproduce realistic pin-hole contact conditions. This choice matters because an optimizer driven by artificial stress peaks will make bad decisions—thickening plates in the wrong places or distorting geometry to relieve stresses that would never occur in service. By distributing the load realistically across the bearing surface of each hinge bore, the model keeps the stress predictions meaningful for the actual contact mechanics of a pinned joint.

The work is a preprint, released on 7 September 2026 and open for community discussion until 18 October 2026, and referees have already raised substantive questions that will shape the final version. Among the issues flagged are the consistency of some optimized variable values with their stated constraints, the justification for selecting the 1-meter excavation condition as representative of the most critical case, the sufficiency of 300 training samples for the surrogate models given the number of design variables, and the absence of fatigue-related indicators even though the paper emphasizes long-term cyclic impact loading. The authors have responded constructively, noting that empirical weighting combined with comparative analysis of multiple weight combinations was used, and that the manuscript will be comprehensively revised in response to the comments.

Even in preprint form, the study offers a template for how heavy equipment design may evolve. Excavator attachments sit at the intersection of kinematics, hydraulics, and structural mechanics, and the failure modes that matter—fatigue cracks initiating at welded hinge lugs, buckling under peak resistance, excessive deflection—depend on all three domains at once. The collaborative optimization framework demonstrated here, pairing test-identified load cases with surrogate-accelerated multi-objective search, provides a reference workflow that other attachment designers can adapt, whether for buckets, booms, or the arms of loaders and backhoes. As computational tools mature and surrogate modeling becomes routine, the era of optimizing one discipline at a time may be drawing to a close, replaced by designs in which mechanism and structure are tuned together from the start.

Subject of Research: Mechanism-structure collaborative optimization of an excavator stick for hard-soil digging conditions

Article Title: Mechanism‑Structure Collaborative Optimization Design of Excavator Stick under Hard‑Soil Conditions

Article References: Ren, K., Ren, Z., Zhang, H., Zhang, Y., Chen, Y., & Liu, R. (2026). Mechanism‑Structure Collaborative Optimization Design of Excavator Stick under Hard‑Soil Conditions. https://doi.org/10.5194/ms-2026-167

Image Credits: AI Generated

DOI: 10.5194/ms-2026-167

Keywords: excavator, stick, collaborative optimization, Kriging surrogate model, NSGA-II, von Mises stress, hard soil, multidisciplinary design optimization, structural lightweighting, digging force, finite element analysis, Mechanical Sciences

The research, published in the International Journal of Intelligent Robotics and Applications, focuses on a cable-driven hyper-redundant robot, a class of machines defined by a very high number of degrees of freedom. Hyper-redundant manipulators are prized for their manoeuvrability in confined and challenging environments, from inspection ducts to maintenance scenarios where conventional rigid arms cannot reach. But that flexibility comes at a price. Because each joint can move independently, the space of feasible postures is vast, and each posture produces a different distribution of internal loads. The mapping from configuration to peak stress is highly nonlinear and can only be evaluated reliably through computationally expensive simulations, making an exhaustive search impractical even for a five-dimensional problem.

The core of the new workflow treats worst-case stress identification as a black-box maximisation problem over the robot’s configuration space. For any candidate pose, a recursive statics model first computes the forces and moments that the rest of the manipulator transmits to the component of interest, in this case the robot’s first link. Cable tensions, gravity and the geometry of the cable routing all enter this equilibrium calculation, which is solved sequentially from the distal end of the arm toward the base. The resulting interface reactions then serve as boundary conditions for a finite element analysis of the link, performed in ANSYS, which returns the peak Von Mises stress, the standard scalar measure engineers use to gauge how close a material is to failure.

Running that pipeline for every conceivable pose would be prohibitively slow, so the researchers built a surrogate model that learns the relationship between configuration and stress. The surrogate is a Gaussian process, a probabilistic model that not only predicts the stress at any pose but also quantifies how confident it is in that prediction. The team used an ARD Matérn 5/2 kernel, a mathematical structure whose per-dimension length scales reveal which joint variables matter most to the stress response, and trained it on an initial set of 200 poses generated by Latin hypercube sampling across the joint limits of minus 45 to plus 45 degrees. All inputs and outputs were z-score standardised before training to keep the numerics well conditioned.

Bayesian optimisation then took over. The algorithm repeatedly asks where the next finite element simulation should be spent, using an acquisition function called Expected Improvement that balances two competing urges: exploiting regions where the model already predicts high stress, and exploring regions where its uncertainty is large, since a hidden danger may lurk there. Rather than evaluating one pose at a time, the framework uses batch Expected Improvement, proposing ten configurations per iteration that are evaluated in parallel. Because the batch criterion accounts for correlations between the proposed points, it penalises redundant, near-identical samples and rewards diverse batches that jointly cover both promising and uncertain regions of the configuration space.

The optimisation converged after just six Bayesian iterations, corresponding to 250 evaluated configurations in total. The identified worst-case pose produced a peak stress of 14.23 megapascals, an improvement of 1.41 percent over the best configuration found in the initial random design. That modest gain is itself informative: the space-filling initial design had already sampled near the critical region, and the Bayesian loop refined and confirmed the most demanding neighbourhood rather than drifting elsewhere. The Expected Improvement values attached to proposed batches fell from 16.2 percent of the normalised output range at the first iteration to 0.21 percent at convergence, indicating genuine saturation rather than a premature stop.

Two further checks strengthened confidence that nothing dangerous had been missed. When the researchers optimised the final Gaussian process surrogate itself and validated its predicted optimum in ANSYS, the resulting configuration, although geometrically distinct from the observed worst case, produced approximately 14.2 megapascals, nearly identical to the maximum found during the loop. This suggests the stress-critical condition is not an isolated spike but a neighbourhood of similarly punishing poses, exactly the kind of insight engineers need when planning real trajectories. On an independent test set of 100 previously unseen configurations, the surrogate achieved a mean absolute error of 0.13 megapascals and a coefficient of determination of 0.992, and in the top decile of the stress distribution its uncertainty estimates proved slightly conservative, with no false negatives at either the 90th or 95th percentile danger thresholds.

The authors are careful to frame the case study as methodological rather than as a certification of the hardware. The first link they analysed is intentionally overdesigned, so the absolute stress values remain far below the 45 megapascal yield strength of the PA12 polymer from which the link is made. The load cases considered arise from self-weight, cable preload and configuration-dependent internal load transfer, with no external payload applied at the end effector, and the analysis is quasi-static, omitting inertial effects and distributed cable friction. Within those assumptions, however, the framework demonstrates something broadly useful: it can tell designers whether substantially higher stresses are likely to remain hidden in unexplored regions of the configuration space, and it can do so with a limited simulation budget.

The practical implications extend beyond design verification. Because the surrogate is trained directly on robot configurations, the same workflow can be used during operation to screen trajectories, flagging or excluding poses associated with elevated structural demand before the robot ever attempts them. This stress-aware trajectory screening could be particularly valuable for hyper-redundant and snake-like robots deployed in inspection and maintenance, where the machine’s posture changes continuously and a single unlucky configuration could compromise the mission. The authors also note that the approach could be extended to multiple links, alternative structural metrics such as fatigue, or constrained multi-objective formulations, and that nonlinear material regimes would require discontinuity-aware kernels or additional regularisation.

What makes the work resonate beyond robotics is its demonstration of a broader principle: when every evaluation is expensive, intelligence about where to look next is worth more than raw computational power. Bayesian optimisation with Gaussian process surrogates has already transformed topology optimisation, aerospace design and co-design of robot mechanics and behaviour, and this study extends that logic into the configuration space of the robots themselves. As machines grow more articulated and their design spaces explode in size, the ability to map where danger concentrates, with a few hundred simulations instead of millions, may become as important to robot safety as the strength of any individual part.

Subject of Research: Bayesian optimisation for identifying stress-critical robot poses in a cable-driven hyper-redundant manipulator

Article Title: Configuration-space Bayesian optimisation for stress-critical pose identification on a hyper-redundant robot

Article References: Poka, A., Ludovico, D., Verme, L. D. M. C. D., Manara, F., Canali, C., Berselli, G., & Caldwell, D. G. (2026). Configuration-space Bayesian optimisation for stress-critical pose identification on a hyper-redundant robot. International Journal of Intelligent Robotics and Applications. https://doi.org/10.1007/s41315-026-00586-z

Image Credits: AI Generated

DOI: 10.1007/s41315-026-00586-z

Keywords: Bayesian optimisation, hyper-redundant robots, Gaussian process, finite element analysis, Von Mises stress, surrogate modelling, cable-driven manipulator, robotics, structural design, worst-case stress, Expected Improvement, configuration space