Soil salinization restricts crop production and complicates field management in arid farming systems. We assessed low-cost red-green-blue and near-infrared sensing across three contiguous management zones assigned to deep vertical rotary tillage depths of 40, 55, and 70 cm in Bachu County, Xinjiang, China. The dataset contained 90 paired sampling points, with 30 spatial subsamples in each zone. Because each depth was represented by one zone, the design had no independent plot-level biological replication and zone contrasts were interpreted descriptively. We compared support vector regression and extreme gradient boosting after feature screening by the maximal information coefficient, Pearson correlation, or grey relational analysis. Hyperparameters were tuned by conventional random search, the Sparrow Search Algorithm, or the Grey Wolf Optimizer within repeated nested cross-validation. Mean soil total salt contents were 12.12, 14.63, and 7.24 g kg⁻¹ in the 40, 55, and 70 cm zones, respectively. Welch’s analysis of variance and Games–Howell comparisons indicated that the 70 cm zone differed from the other two zones, whereas the 40 and 55 cm zones did not differ. These point-level tests describe spatial contrasts and do not establish a causal tillage-depth effect. Using field-derived RGB-NIR predictors, the best screened and conventionally tuned extreme-gradient-boosting model achieved a pooled out-of-fold coefficient of determination of 0.694 and a root mean square error of 3.832 g kg⁻¹. Its paired outer-fold error was lower than that of the all-feature default baseline after multiplicity correction, while the two swarm optimizers did not significantly improve on conventional tuning under the matched evaluation budget. The results support field-derived RGB-NIR sensing as a practical aid for within-field monitoring of soil total salt content, while non-spatial cross-validation may overestimate performance at spatially independent locations and external validation remains necessary.

Apocynum venetum has clustered, highly branched stems with strong phloem fiber adhesion, which makes mechanical harvesting prone to entanglement, grip slippage, conveyance blockage, and uneven, high stubble. Existing cutting and bundling machines do not match this morphology well and cannot meet both low-stubble and anti-clogging requirements. To address these problems, an integrated Apocynum venetum cutting-bundling harvester was developed that combines a reciprocating cutting mechanism, stem-folding assembly, vertical anti-clogging conveyor at the cutting table, height-adjustable crop-plate compaction device, and cord-based bundling mechanism, which was supported by a combined navigation system integrating BeiDou satellite navigation and inertial navigation. Field trials in Xinjiang demonstrated that the prototype achieved a mean stubble height of 12.4 cm, a blockage incidence of 0.30 events per hour, a bundling success rate of 98\%, a fiber damage rate of 0.8\% and a harvesting efficiency of 0.0537 ha h\(^{-1}\), meeting ISO 11279:2021 performance thresholds. The coordinated action of the cutting, folding, conveying, and bundling subsystems improved the throughput and maintained stable machine operation under clustered growth conditions. These results indicate that the developed prototype harvester is technically feasible and agronomically suitable for large-scale Apocynum venetum production. The intelligent Apocynum baling harvester can enhance the automatic driving functionality of agricultural machinery, improve the operational quality of agricultural machinery navigation, and thereby comprehensively increase harvesting efficiency.

The absence of discrete element contact parameters for suitable harvest rapeseed pods hinders accurate DEM simulations. This study presents the first systematic calibration of discrete element contact parameters for “Ningza 158” rapeseed pods using physical and simulation experiments. Experimentally determined properties included external dimensions, density, pod–steel coefficients (restitution, static friction, rolling friction), and pod–pod restitution. Using the measured angle of repose as the objective, significant parameters were screened, optimized, and modeled via Plackett–Burman, steepest climbing, and central composite response surface tests. The optimal parameter combination was then validated against the angle of repose, completing the calibration. The results showed that the density, Poisson’s ratio, and shear modulus of rapeseed pods at the suitable harvest period were 346.63 kg·m⁻³, 0.35, and 10.5 MPa, respectively. The coefficients of restitution, static friction, and rolling friction for pod–steel and pod–pod were 0.2994, 0.4206, 0.0435, 0.1024, 0.3110, and 0.0380, respectively. Under this simulation parameter combination, the angle of repose obtained from the simulation was 71.23°, yielding a relative error of 2.12% compared with the actual value of 69.75°. Note that this validation is limited to static accumulation behavior. The calibrated parameters can be a basis for future dynamic validation studies on harvest loss mechanisms, provided additional dynamic validation is performed.